You are on page 1of 198

MEM REX

3672
Storage Control Unit
Theory of Operations Manual

3672.21-00
3672.21-0001

Publication Number 3672.21-00


and Publication Bulletin Number 3672.21-0001
Part Number

5,

j, D 4065

November 1975 edition


The information in this publication is subject to periodic revisions and to
updates by Publications Bulletins. A bulletin is identified as Publication
Number 3672.21-00XX, where "XX" equals "01" for the first bulletin
released.
To request copies of this publication or any bulletin, send a Supplies
Requisition (form number 7020) using the publication number to: Memorex
Corporation, Marketing Distribution Center, San Tomas at Central Ex.
pressway, Santa Clara, California 95052.
A readers' comments form is provided at the back of this publication. If the
form has been removed, please send comments to Memorex Corporation,
Department 6940, MIS 14-07, San Tomas at Central Expressway, Santa
Clara, California 95052.

1974, 1975, MEMOREX CORPORATION

PREFACE
This manual provides the Memorex Field Engineer (FE) with detailed operating theory about the
3672 Storage Control Unit, used in 3672-based disc storage subsystems with the 3673 Disc Drive
Controller and 3670/3675 Disc Drive Modules. It is intended for use by the FE during training as
essential course material, and during servicing as an aid in identifying equipment difficulties. The
manual consists of six sections:
Section 1 -

Describes briefly the overall functions performed by the 3672 Storage Control
Unit and the 3672-based subsystem.

Section 2 -

Describes commands executed by the 3672. Includes command format and


summary information, and a detailed flow diagram of each command.

Section 3 -

Presents details about the 3672/channel interface.

Section 4 -

Contains a description of the microinstructions and microprograms used to


execute commands.

Section 5 -

Describes operation of the 650 Flexible Disc File, used to store operating and
diagnostic microprograms executed by the 3672.

Section 6 -

Describes the detailed operation of the 3672 logic.

Maximum benefit of this manual is achieved when used with th,e 3672 Storage Control Unit Logic
Diagram Manual PIN 308312. The alphanumeric numbers which appear in each functional block
of the block diagrams in Section 6 refer to corresponding pages of logic in the Logic Diagram
Manual.
Other manuals that support the 3672 Storage Control Unit which may be of use to the FE are listed
below:
3672.22-00 - 3672 Storage Control Unit Installation Manual
3672.20-00 -

3672 Storage Control Unit Maintenance Manual

3672.50-01 - 3672 Storage Control Unit Microdiagnostics Reference Manual (two volumes)

TABLE OF CONTENTS

Section

I
3672.21 -0001-10175

Page

GENERAL DESCRIPTION

1-1

1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.7.1
1.7.2
1.7.3
1.8

1-1
1-1
1-1
1-1
1-3
1-3
1-3
1-3
1-4
1-4
1-6

SCOPE
SUBSYSTEM CHARACTERISTICS
STANDARD FEATURES
OPTIONAL FEATURES
COMPATIBILTY
SPECIFICATIONS SUMMARY
UNIT CHARACTERISTICS
3672 Storage Control Unit
3673 Disc Drive Controller
3670 (3675) Disc Drive Module
DATA INTERFACE REQUIREMENTS

COMMANDS

2-1

2.1
2.2
2.3
2.3.1
2.3.2

2-1
2-1
2-1
2-1
2-1
2-1
2-1
2-1
2-1
2-1
2-1
2-3
2-3
2-4
2-4
2-4
2-4
2-4
2-4
2-5
2-5
2-5
2-5
2-5
2-5
2-5
2-5
2-7
2-7

2.3.3
2.3.4
2.3.5
2.3.5.1
2.3.5.2
2.3.5.3
2.3.5.4
2.3.5.5
2.3.6
2.4
2.4.1
2.4.1.1
2.4.1.2
2.4.1.3
2.4.1.4
2.4.1.5
2.4.2

GENERAL
COMMAND SUMMARY
INSTRUCTION AND WORD FORMAT
Introduction
1/0 Instructions
START 1/0
START 1/0 FAST RELEASE
TEST I/O
HALT 1/0
HALT DEVICE
Channel Address Word
Channel Command Word
Channel Status Word
Status Presentation
Initial Status
Ending Status
Pending Status
Contingent Connection
Program Status Word
CONTROL COMMANDS
Operational Block Diagram
System
System Interface
Storage Control Unit
Drive Controller
Disc Drive Module
Command Descriptions
NO-OP
SEEK

Section

Page
SEEK CYLINDER
SEEK HEAD
SPACE COUNT
RECALIBRATE
RESTORE
SET FILE MASK
DIAGNOSTIC LOAD
DIAGNOSTIC WRITE'

2.5
2.5.1
2.5.2
2.5.3
2.5.4
2.5.5

2.5.6
2.5.6.1
2.5.6.2
2.5.6.3
2.5.6.4
2.5.6.5
2.5.6.6
2.6
2.6.1
2.6.1.1
2.6.1.2
2.6.1.3
2.6.1.4
2.6.1.5
2.6.1.6
2.6.1.7
2.6.1.8
2.6.2

SENSE COMMANDS
Sense Byte Information
Sense Control Block
Sense Byte Bit Definitions
Assembling Sense Bytes 8 through 23
Command Descriptions
SENSE 1/0
TEST I/O
READ AND RESET BUFFERED LOG
READ DIAGNOSTIC STATUS 1
DEVICE RESERVE
DEVICE RELEASE
Transfer In-Line Diagnostics to SCU
Introduction
System
Channel
Channel Interface
Storage Control Unit (SCU)
Control Storage
READ COMMANDS
Operation Block Diagram
System
Channel
Channel Interface
Microprocessor
Controller
Read Circuit
Read/Write Head
Pack
Command Descriptions
READ DATA
READ KEY AND DA1A
READ COUNT, KEY, AND DATA
READ RECORD ZERO
READ COUNT
READ HOME ADDRESS
READ IPL
READ SECTOR

2-8
2-8
2-9
2-10
2-11
2-12
2-14
2-15
2-16
2-16
2-16
2-16
2-16
2-16
2-21
2-22
2-23
2-24
2-25
2-25
2-26
2-26
2-26
2-26
2-26
2-26
2-26
2-27
2-27
2-27
2-27
2-27
2-27
2-27
2-27
2-27
2-27
2-27
2-28
2-28
2-29
2-32
2-32
2-33
2-36
2-36

Page

Section

2.6.2
2.6.4
2.6.4.1
2.6.4.2
2.6.4.3
2.6.4.4
2.6.4.5
2.6.4.6
2.6.4.7
2.6.4.8
2.7
2.7.1
2.7.1.1
2.7.1.2
2.7.1.3
2.7.1.4
2.7.1.5
2.7.1.6
2.7.1.7
2.7.1.8
2.7.2

~.7.3

2.8

2.9
2.10
2.11
2.111
2.11.1.1
2.11.1.2
2.11.1.3
2.11.1.4
2.11.1.5
2.11.2
2.11.3
2.11.4
2.11.5
2.11.6

Read Data Transfer


Read Diagnostic to System
Introduction
System
Channel
Channel Controls
650 Drive
650 Interface
Control Storage'
Control Unit
WRITE COMMANDS
Operation 6Jock Diagram
System
Channel
Channel Interface
Microprocessor
Controller
Write Circuits
Read/Write Head
Pack
Command Descriptions
WRITE DATA
WRITE KEY AND DATA
WRITE COUNT, KEY, AND DATA
WRITE SPECIAL COUNT, KEY, AND DATA
WRITE RECORD ZERO
WRITE HOME ADDRESS
ERASE
Write Data Transfer
SEARCH COMMANDS
SEARCH HOME ADDRESS EQUAL
SEARCH ID EQUAL
SEARCH 10 HIGH
SEARCH 10 EQUAL OR HIGH
SEARCH KEY EQUAL
SEARCH KEY HIGH
SEARCH KEY EQUAL OR HIGH
SELECTIVE RESET
SYSTEM RESET
COMMAND RETRY
Check and Error-Initiated Retry
Data Checks
Sync Check
AM Check
Physical 10 Error
Seek Incomplete Error
Defective or Alternate Track
Command and Data Overrun
Padding
Write Offset
Invalid Count Field Sync Detected

2-38
2-38
2-38
2-38
2-38
2-38
2-38
2-38
2-38
2-38
2-42
2-42
2-42
2-42
2-42
2-42
2-43
2-42
2-42
2-42
2-44
2-43
2-43
2-46
2-47
2-48
2-49
2-50
2-51
2-54
2-54
2-56
2-56
2-57
2-59
2-59
2-60
2-62
2-63
2-64
2-74
2-64
2-64
2-64
2-66
2-64
2-64
2-64
2-64
2-64
2-64

Section

Page

2.11.7
2.11.8
2.12
2.12.1
2.12.2
2.12.3
2.12.4
2.13
2.14
2.14.1
2.14.2
2.14.3
2.15

Index Detected in Data Field


Command Retry Flowcharts
ROTATIONAL POSITION SENSING
Introduction
Read Sector
Set Sector
Disc and Track Layout
MULTITRACK OPERATION
OVERFLOW RECORD
Introduction
Formatting
Processing
MULTICHANNEL OPERATION

CHANNEL INTERFACE DESCRIPTION


3.1
3.2
3.2.1
3.2.2
3.3
3.3.1
3.3.1.1
3.3.1.2
3.3.1.3
3.3.1.4
3.3.1.5
3.3.1.6
3.3.2
3.3.2.1
3.3.2.2
3.3.3
3.3.3.1
3.3.3.2
3.3.4
3.3.4.1
3.3.4.2
3.4
3.4.1
3.4.2
3.4.2.1
3.4.2.2
3.4.2.3
3.4.2.4
3.5
3.5.1
3.5.2
3.5.3
3.5.4
3.5.5

GENERAL
CHANNEL INTERFACE LINES DESCRIPTION
Inbound Lines
Outbound Lines
INTERFACE SEQUENCES
Overall Descriptions
Initial Selection Sequence
Data Transmission Sequence
Ending Sequence
SCU Busy Response
SCU-Initiated Sequence
Immediate Command Sequence
Initial Selection Sequence
Simplified Sequence
Detailed Sequence
Ending Sequence
Simplified Sequence
Detailed Sequence
Polling Sequence and Status Presentation
Simplified Sequence
Detailed Sequence
CHANNEL OPERATION
Selector Channel Operation
Block Multiplexer Operation
Introduction
Rotational Position Sensing
Disconnected Command Chaining
Multiplexer Channel Operation
SEQUENCE CONTROLS
Command Chaining
Data Chaining
Branching in Channel Programs
Unit Selection and Device Addressing
Stack Status

2-64
2-64
2-77
2-77
2-77
2-77
2-78
2-79
2-80
2-80
2-80
2-80
2-81

3-1
3-1
3-2
3-2
3-2
3-4
3-4
3-4
3-4
3-4
3-4
3-4
3-4
3-4
3-4
3-6
3-6
3-6
3-7
3-7
3-7
3-7
3-8
3-8
3-8
3-8
3-8
3-8
3-8
3-13
3-13
3-13
3-13
3-13
3-13

vi

Section

Page

3.5.6
3.57
3.5.8
3.5.9
3.5.10
3.5.11
3.5.12
3.5.13
3.5.14
3.5.15
3.5.16
3.5.17
3.6
3.6.1
3.6.2
3.6.2.1
3.6.2.2
3.6.2.3
3.6.2.4
3.6.2.5
3.6.2.6
3.6.2.7
3.7
3.7.1
3.7.2
3.7.3
3.8
3.8.1
3.8.2
3.9
3.9.1

3.11
3.11.1
3.11.2
3.11.3
3.11.4
3.11.5
3.11.6
3.12
3.12.1
3.12.2
3.12.2.1
3.12.2.2
3.12.2.3
3.12.2.4
3.12.2.5
3.12.3
3.12.3.1

3672.21-0001-10175

Suppress Status
Disconnect In
Interface Disconnect
Selective Reset
Command Retry
Proceed
Stop (or Truncation)
Suppress Data
Data Acceptance
Data Ready
Status Acceptance
System Reset
STATUS CONDITIONS
Overview
Description of Status Conditions
Status Modifier
Control Unit End
Busy
Channel End
Device End
Unit Check
Unit Exception
SENSE CONDITIONS
Overview
Conditions Indicated by Bits of Sense Byte 0
Conditions Indicated by Bits of Sense Bytes 1 through 23
READ DATA
Description
Flow Diagram
WRITE DATA TRANSFER
Description
Flow Diagram
SCU-INITIATED CHECK 1 ERROR CONTROL
SEQUENCE (pOLLING). SIMPLIFIED
MULTICHANNEL SWITCH OPERATION
Channel Selection Switch
Instantaneous Connection
Long Connection
Device Status
Addressing
Resets
2860 SELECTOR CHANNEL ATTACHMENT FEATURE
Description
Functional Characteristics
General
Channel Commands
Disconnected Command Chaining
Error Recovery
Non-RPS Mode Operation
1/0 Programming
Channel Programs

3-13
3-13
3-13
3-13
3-14
3-14
3-14
3-14
3-14
3-14
3-14
3-14
3-15
3-15
3-15
3-15
3-15
3-15
3-15
3-15
3-16
3-16
3-17
3-17
3-17
3-17
3-18
3-18
3-19
3-23
3-23
3-24
3-29
3-30
3-30
3-30
3-30
3-30
3-30
3-30
3-32
3-32
3-32
3-32
3-33
3-34
3-34
3-35
3-35
3-35

Section

Page

I
4

3.12.3.2
3.12.3.3

MICROPROGRAM

4.1
4.1.1
4.1.2
4.1.3
4.1.3.1
4.1.3.2
4.1.3.3
4.1.3.4
4.1.4
4.1.5
4.2
4.2.1
4.2.2
4.2.3
5

Status Condition Exceptions


1/0 Interrupts

MICROINSTRUCTION
General
Microinstruction Word Description
Microinstruction Format
Format 0 Microinstructions
Format 1 Microinstructions
Format 2 Microinstructions
Format 3 Microinstructions
Microinstruction Field Definitions
Field Translation
MICROPROGRAM
Microprogram Block Format
Example Microprogram
Microprogram Routine .organization

FLEXIBLE DISC FILE

5.1
5.2
5.2.1
5.2.2
5.2.3
5.3
5.3.1
5.3.2
5.3.3
5.3.4
5.4
5.4.1
5.4.2
5.4.3
5.4.3.2
5.4.3.2
5.4.4
5.4.4.1
5.4.4.2
5.4.4.3
5.4.5
5.4.5.1
5.4.5.2
5.4.5.3
5.5
5.5.1
5.5.2
5.5.3

GENERAL DESCRIPTION
SPECIFICATIONS
Machine Characteristics
Operating Capabilities
Power Requirements
SPECIAL PRECAUTIONS
Cartridge Loading
Disc Interchangeability
Physical Damage
Safety
PRINCIPLES OF OPERATION
General Operation
Functional Assemblies
Logic and Analog Functional Descriptions
Control and Status Logic
Read Logic
Functional Operation
Initialization Phase
Track Access Phase
Read Phase
Interface Signal Description
Control
Data
Power
CONTROL CIRCUITRY DESCRIPTION
Functional Description
Printed Circuit Board
Harness Assembly

3-35
3-35
4-1
4-1
4-1
4-1
4-1
4-1
4-1
4-1
4-1
4-1
4-1
4-5
4-5
4-5
4-5
5-1
5-1
5-2
5-2
5-2
5-2
5-3
5-3
5-3
5-3
5-3
5-4
5-4
5-5
5-6
5-6
5-6
5-6
5-6
5-6
5-7
5-7
5-8
5-9
5-9
5-10
5-10
5-10
5-10

vii

Section

Page

STORAGE CONTROL UNIT OPERATIONS

6.1
6.2
6.2.1
6.2.2
6.2.2.1
6.2.2.2
6.2.2.3
6.2.2.4
6.2.2.5
6.2.3
6.2.3.1
6.2.3.2
6.2.3.3
6.2.4
6.2.4.1
6.2.4.2
6.3
6.3.1
6.3.2
6.3.2.1
6.3.2.2
6.3.2.3
6.3.3
6.3.3.1
6.3.3.2
6.3.3.3
6.3.3.4
6.3.3.5
6.3.4

GENERAL DESCRIPTION
CHANNEL INTERFACE DESCRIPTION
General
Logic Description
Address Compare Logic
Select Logic
Priority Logic
Multiplexer IDemultiplexer Circuits
Registers
Data Transfer Organization
General
Read Operations
Write Operations
Channel Interface Branch Conditions
Channel Buffer Ready (CBR)
Queue Empty
MICROPROCESSOR DESCRIPTION
Microprocessor General Description
Microinstructions
Descriptions and Formats
Microinstruction Field Translation and Branching
Microinstruction Execution Example
Writable Control Storage
General
Reading From WCS
Writing Into WCS
Error Correction Code
WCS Expansion
Data Paths

6-1
6-2
6-2
6-2
6-2
6-2
6-2
6-2
6-2
6-2
6-5
6-5
6-5
6-5
6-6
6-6
6-6
6-7
6-7
6-7
6-7
6-7
6-8
6-8
6-8
6-8
6-9
6-9
6-12
6-12

Section

Page

6.3.4.1
6.3.4.2
6.3.4.3
6.3.5
6.3.5.1
6.3.5.2
6.3.5.3
6.3.5.4
6.3.5.5
6.3.6
6.3.6.1
6.3.6.2
6.3.6.3
6.3.7
6.4
6.4.1
6.4.2
6.4.2.1
6.4.2.2
6.4.3
6.5
6.5.1
6.5.2
6.6
6.6.1
6.6.2
6.6.3
6.6.4

Buses
Registers
Arithmetic Logic Unit
Buffer Storage
General
Description
Buffer Control Commands
Error Correction Code
Buffer Storage Allocation
Flexible Disc Interface
General
Functional Description
IPL Flow Chart
Microprocessor Timing
CONTROLLER INTERFACE
General
Interface Signals
SCU to Controller Signals
Controller to SCU Signals
Register Definitions
FE INTERFACE
General
Functional Description
CHECK-l ERRORS
Error Detection and Sequences
Machine Check (MCK) Register
Check-l Error Display
Reset of Check-l Error Conditions

GLOSSARY OF TERMS

6-12
6-14
6-16
6-16
6-16
6-16
6-18
6-18
6-19
6-19
6-19
6-19
6-23
6-23
6-30
6-30
6-30
6-31
6-31
6-31
6-33
6-33
6-33
6-34
6-34
6-34
6-34
6-34

G-l

viii

LIST OF ILLUSTRATIONS

Figure

1-1
1-2
1-3
2-1
2-1A
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
2-14
2-15
2-16
2-17
2-18
2-19
2-20
2-21
2-22
2-23
2-24
2-25
2-26
2-27
2-28
2-29
2-30
2-3.1
2-32
2-33
2-34
2-35

2-36
2-37
2-38
2-39
2-40

3672-Based Disc Storage Subsystem Components


3672-Based Disc Storage Subsystem Configurations
Disc Storage Subsystem Operational Block Diagram
Seek Operation Block Diagram
Initial Selection Flow Diagram
No-Op Flow Diagram
Seek Flow Diagram
Space Count Flow Diagram
Recalibrate Flow Diagram
Restore Flow Diagram
Set File Mask Flow Diagram
Set Sector Flow Diagram
Diagnostic Load Flow Diagram
Diagnostic Write Flow Diagram
Sense Byte Definitions
Sense 110 Flow Diagram
Test I/O Flow Diagram
Read and Reset Buffer Log Flow Diagram
Read Diagnostic Status 1 Flow Diagram
Device Reserve and Device Release Flow Diagrams
Transfer In-Line Diagnostics to SCU Block Diagram
Read Operation Block Diagram
Read Data, Read Key and Data, and Read Count, Key, and Data Flow Diagrams
Read Record Zp.ro, Read Count, and Read Home Address Flow Diagram
Read IPL and Read Sector Flow Diagrams
Read Data Transfer Flow and Block Diagram
Read Loop Microprogram Subroutine
Read Diagnostic to System Operation
Write Operation Clock Diagram
Write Count, Key, and Data; Write Key and Data; Write Data Flow Diagrams
Write Special Count, Key, and Data Flow Diagram
Write Record Zero Flow Diagram
Write Home Address Flow Diagram
Erase Flow Diagram
Write Data Transfer
Write Loop Microprogram Subroutine
Search Home Address Equal Flow Diagram
Search ID Equal, Search ID High, and Search ID Equal or
High Flow Diagram
Search Key Equal, Search Key High, and Search Key Equal or High Flow
Diagram
Selective Reset Flow Diagram
System Reset Flow Diagram
Command Retry, Data, Sync, and AM Checks
Command Retry, Physical ID and Seek Errors
Command Retry, Defective Track

3672.21-0001-10 '75

Page

Figure

1-2
1-3
1-5
2-5
2-5A
2-7
2-7
2-9
2-10
2-11
2-12
2-13
2-14
2-15
2-16
2-21
2-22
2-23
2-24
2-25
2-26
2-27
2-30
2-34
2-37
2-39
2-40
2-41
2-42
2-44
2-47
2-48
2-49
2-50
2-51
2-53
2-55

2-41
2-42
2-43
2-44
2-45

2-58
2-60
2-62
2-63
2-65
2-66
2-68

I
I

2-46
2-47
2-48
2-49
2-50
2-51
3-1
3-2
3-3
3-4
3-5
3-6
3-7
3-8
3-9
3-10
3-11
3-12
3-13
4-1
4-2
4-3
4-4
4-5
5-1
5-2
5-3
5-4
5-5
5-6
5-7
5-8
5-9
5-10
5-11
5-12
5-13
6-1
6-2
6-3

Page
Command Retry, Alternate Track
Command Retry, Data and Command Overruns
Command Retry, Padding
Command Retry, Write Offset
Command Retry, Invalid Count Field Sync Detected and
Index Detected in Data Field
Read/Set Sector Operation
Track Sector Layout
Typical Track Field Layout
Multitrack Operation
Overflow Record Operation
Multichannel Operation Block Diagram
Serial Connection of SELECT IN/SELECT OUT
Selector Channel Operation. Data Transfer Sequences
Selector Channel Operation, Miscellaneous Sequences
Disc Track Sector Format
Disconnect Command Chaining Block Diagram and Examples
Block Multiplexer Channel Operation (1 of 2)
Read Data Transfer Block Diagram
Read Data Transfer Flow Diagram (1 of 3)
Write Data Transfer Block Diagram
Write Data Transfer Flow Diagram (1 of 4)
Check-l Error Controlling Sequence Flowchart and Logic Diagram
Multi-Channel Switching Logic
Set RPS Flow Diagram
Microinstruction Formats
Microinstruction Expansion
Microprogram Block Formats
Microprogram Format Example
Microprogram Routine Organization
Memorex 650 Flexible Disc File
Flexible Disc Cartridge
Cartridge Loading
Flexible Disc File/Microprocessor Interface
Flexible Disc File Functional Block Diagram
Head Loading Mechanical Assembly
Disc Cartridge and Disc Configuration
Initialization Sequence
Read Sequence
Read Data Waveforms
Flexible Disc File Interface Diagram
Control Circuitry Block Diagram
Flexible Disc File PCB
3672 SCU Block Diagram
Channel Interface Block Diagram
Read Data Transfer Block Diagram

2-70
2-72
2-74
2-74
2-75
2-77
2-78
2-78
2-79
2-80
2-81
3-3
3-5
3-9
3-10
3-10
3-11
3-18
3-20
3-23
3-25
3-29
3-31
3-33
4-2
4-4
4-6
4-7
4-8
5-1
5-1
5-3
5-4
5-5
5-5
5-5
5-6
5-7
5-7
5-8
5-10
5-10
6-1
6-3
6-5

ix

Figure
6-4
6-5
6-6
6-7
6-8
6-9
6-10
6-11
6-12
6-13
6-14
6-15

Figure

Page
Write Data Transfer Block Diagram
Microprocessor Block Diagram
Microinstruction Expansion
Microinstruction Translation
Branch Multiplexing Circuits
Microinstruction Execution Example
WCS Block Diagram
RAM Board Column Select
WCS Extension Block Diagram
Microprocessor Buses and Registers
Buses and ALU Block Diagram
Buffer Storage Block Diagram

6-6
6-7
6-7
6-8
6-9
6-10
6-11
6-12
6 e 12
6-13
6-17
6-18

6-16
6-16A
6-17
6-18
6-19
6-20
6-21
6-22
6-23
6-24
6-25
6-26
6-26

Page
Buffer Storage Allocation and Contents
Buffer Storage Byte Definitions
Flexible Disc Interface Block Diagram
FD Track Format
IPL Flow Chart
Microprocessor Timing Logic
Nominal Microprocessor Timing
Detailed Microprocessor Clock Timing
Detailed Microprocessor SCU Addressing and ALU Path Timing
Controller Interface Block Diagram
Storage Control Interface Line Drivers and Receivers
SCU to Controller Interface Timing
Controller to SCU Interface Timing (Data Transfers)

6-20
6-20A
6-21
6-23
6-24
6-26
6-27
6-28
6-29
6-30
6-30
6-31
6-31

LIST OF TABLES

Table

Page

Table

2-1
2-2
2-3
2-4
2-5
2-6
4-1
5-1
5-2

2-2
2-16
2-17
2-18
2-19
2-81
4-3
5-8
5-8

5-3
5-4

Command Summary
Sense Control Block
Sense Byte 0-7 Bit Definitions
Message Table Formats
Sense Bytes 8-23 Bit Definitions
Device Reserve/Release Command Functions
Microinstruction Field Definitions
Input Control Signals
Output Control Signals

3672.21-0001-10175

6-'
6-2
6-3
6-4
6-5

6-6
6-7

Page
Data Signals
Power Requirements
WCS ECC Matrix
Example Data Word Applied to WCS ECC
Check-1 Multiplex Error Conditions
Buffer Storage ECC Matrix
Example Data Word Applied to Buffer Storage ECC Matrix
FDF Multiplexer Output Bits
ECR Bits

5-9
5-9
6-14
6-14
6-16
6-19
6-19
6-23
6-32

SECTION 1. GENERAL DESCRIPTION

1.1 SCOPE
This manual contains operating principles for the
MEMOREX 3672 Storage Control Unit (SCU), which
functions as the primary controlling element in the
MEMOREX 3672-based Disc Storage Subsystem. Since
the SCU exercises primary control in the Subsystem and
participates in all Subsystem operations, this section of
the manual will. describe pertinent characteristics of the
complete 3672-based Subsystem, with particular details
about the 3672 SCU in the following sections.

1.2 SUBSYSTEM CHARACTERISTICS


The MEMOREX 3672-based Disc Storage Subsystem
(Figure 1-1) provides high performance. large capacity.
direct access data storage. The 3672-based Subsystem is
completely
compatible
with
the
IBM
3830-11/
3333/3330 Disc Storage Subsystem and can interface
with IBM System/360 and 370 which utilize a Block
Multiplexer Channel.
The 3672-based Subsystem IS composed of a 3672 SCU
and up to four 3673 DIsc Drive Controllers, each
attached to four 3670 or 3675 Disc Drive Modules
(DDM). The 3675 DDM is a double-density version of the
3670 DDM. Each DDM contains two drive spindles. The
DDMs may be mixed in any combination (Figure 1-2).

The 3672 SCU attaches on the upstream side to an IBM


System/360 or 370 through a Block Multiplexer
Channel. or to a Selector Channel by means of the
optional 2860 Selector Channel Attachment Feature. The
SCU attaches on the downstream side to a maximum of
four 3673 Controllers "through a Controller Interface
(CTL-I). Each Controller, in turn, may attach to as many
as four DDMs. The SCU may be shared between two,
three. or four Central ProceSSing Units (CPUs) by means
of an optional channel switch attachment in the SCU.
and may connect to one to four 3673 Controllers through
an optional string switch feature in the Controllers.
Storage media used in the 3672-based Subsystem is the
MEMOREX Mark X, IBM 3336-1. or equivalent disc pack
for subsystems utilizing 3670 DDMS; or the MEMOREX
Mark Xl. IBM 3336-11, or equivalent disc pack for
subsystems eqUipped with 3675 DDMs.

3672.21-0001-10/75

Removable logical address plugs permit changing the


logical device addresses of the drives within the 3672based Subsystem. A service plug is provided with each
Subsystem to enable offline or inline servicing by the
Memorex Field Engineer (FE).
The 3672-based Subsystem uses IBM System/360 or
370 Channel Control Words (CCWs) to identify the data
operation to be performed, and the addresses of the
3672 SCU. 3673 Controllers, and 367X DDMs that are to
be controlled. The SCU transmits control Signals to the
selected controller and drive control circuitry in logical
sequence and at the proper time to complete the desired
operation.

1.3 STANDARD FEATURES

may be loaded into the WCS for execution. The disc is


small and easily changed; therefore, program updates
can be made easily by changing the disc.

Usage/Error Recording

1.4 OPTIONAL FEATURES


Provides a means of proceSSing records which exceed
track boundaries within a cylinder.

Error Correction

Program Compatibility

Error Recovery
The Subsystem receives, decodes, and interprets
commands from the IBM System/360 (with Block
Multiplexer) or 370 channel. It responds to the same
set of (CCWs) used in IBM programs to operate the
3330 Facility.

Microprograms are stored in Writable Control Storage


(WCS). which is loaded from a flexible disc in the
SCU
Flexible Discs
The flexible discs contain the operating microprogram.
diagnostic programs, and other backup programs that

Data Protection
To protect data, the operator can inhibit write commands by utilizing a READ-ONLY switch on each
drive.

System Disabling
The operator can easily take the 3672-based Subsystem offline (for diagnostic testing or maintenance)

Four-Channel Switch
The four-channel switch is identical to the twochannel switch in operation; however. four independent channels may be connected to the 3672 SCU.
These four independent channels may be connected
with one to four separate CPUs.

Operator/Diagnostic Console
To facilitate operator monitoring and FE maintenance.
each drive within the module employs a separate
Operator /Diagnostic console.

Three-Channel Switch
The three-channel switch is identical to the twochannel switch in operation. except that three independent channels may be connected to the 3672
SCU. The three independent channels may be connected with one to three separate CPUs.

Rotational Position Sensing

Microprogramming

Two-Channel Switch
The two-channel switch provides the capability for the
3672 SCU to be shared by two IBM System/360 or
370 Block Multiplexer Channels. The two channels
may be attached to either the same or different CPu;;. ~
Individual drives attached to the 3672 SCU may be ---reserved for the exclusive use of either of the
channels.

The logical address of any disc drive is easily changed


by changing the unit select plug on the drive operator
control panel.

Rotational position sensing allows the location of a


record through its angular displacement relative to
index and releases the channel during most of the
record search time.

Drive Addressing

Command retry capability of the SCU allows recovery


from subsystem errors and permits the using system
to be available during retry attempts.

Priority
The Memorex FE can assign a priority to each 3672
SCU and its address.

Record Overflow

The SCU maintains a record of statistical data


(number of seeks executed and number of bytes read)
and error information for each drive.

Error correction capability allows the detection and


correction of data errors within the Subsystem.

by setting a switch on the 3672 SCU operator control


panel and ensuring that the CPU enters the Wait
State momentarily.

Tag/Untag Switch
This switch is utilized on SCUs equipped with one of
the multiple channel switch features. When the
switch is in the TAG position. each channel must
accept the Device End Signal resulting from a pack
change or a unit plug change before the channel can

1-1

a. 3672 Storage Control Unit

b. 3673 Disc Drive Controller

c. 3670/3675 Disc Drive Module

Figure 1-1. 3672-Based Disc Storage Subsystem Components

1-2

process data with that device In the UNTAG position.


acceptance of the Device End signal resulting from a
pack change or a unit plug change by a channel will
allow any channel to process data with that device.
3670
3675

This feature allows for attaching up to 32 disc drive


spindles (16 modules) to one 3672 SCU. The WCS
ExpanSion is a prerequisite for this feature.

IBM

SYS 360
SYS 370

WCS Expansion

No. of Recording Discs

Expands WCS to support additional microprogramming. Required for 32-spindle addressing.

J+--'-----.

3672
SCU

19

Tracks Per Surface

404 ~ 7 alternates (3670)


808 + 7 alternates (3675)

Track Capacity

13,030 bytes

Cylinder Capacity

247,570 bytes

Pack Capacity

100,018,280 bytes (3670)


200,036,560 bytes (3675)

Module Capacity

200,036,560 bytes (3670)


400,073,120 bytes (3675)

Data Recording Format

Power Options
A subsystem can be supplied to operate from a threephase 2081230 Vac 60 Hz or 220/380 Vac 50 Hz
power source.

IBM 3330 Compatible

2860 Selector Channel Attachment

In the discussion which follows. refer to the 3672-based


Subsystem block diagram on Figure 1-3.

Permits the 3672-based Subsystem to be attached to


a System/360 selector channel. Enables implementation of the software-provided Disconnected Command
Chaining and Command Retry functions required for

simulated block multiplexer operation with this attachment.


CTLI

1.5 COMPATIBILITY

UP TO
4 CON
TROLLERS
PER SCU

UP TO

L_-"

4 DDM,
(8 DRIVES,
PER CON
TROLLER

LI _ _

The 3672-based Subsystem attaches directly to the IBM


System/370 Block Multiplexer Channels for Models 135
and up, and to System/360 Block Multiplexer Channels
on Model 195. The Subsystem attaches to the selector
channel, via the 2860 Selector Channel Attachment
feature, of IBM System/360 Models 65, 67, 75, 91, and
95.

1.7 UNIT CHARACTERISTICS

1 .7.1 3672 Storage Control Unit


The 3672 SCU is a microprogrammed control unit with
the microprogram stored in WCS. The WCS has its
program loaded during the Power-On sequence from a
flexible disc device (MRX 650) located in the SCU.
In general, the functions included in the SCU are as
follows:

Interfaces with the channel, receives commands from


the channel, interprets and executes them.

Interfaces with 3673 Controllers and 367X DDMs and


controls the operation of each.

Controls the transfer of information between the


Controller and channel.

Checks data and command for validity during all


operations.

Corrects data errors, or sends a correction algorithm


to the channel for data correction, depending on
where the error occured.

1.6 SPECIFICATIONS SUMMARY

Data Retrieval Times

Figure 12. 3672-Based Subsystem Configurations

10

No of Recording Surfaces

CTLI

CHNL

""'-1

32-Spindle Feature

Disc Pack Characteristics

Average Latency

8.33 msec

Track-to-Track Access Time

7 msec (maximum)

Maximum Access Time

50 msec

Average Access Time

27 msec

Data Transfer Rate

806,000 bytes/sec

1-3
3672.21-0001-10/75

internally and then sends the corrected data to the


channel. Upon reorientation to the field in error,
the corrected data is transferred from the buffer to
the channel. During the reorientation time, the
SCU is disconnected from the channel.

Performs diagnostic evaluation of the Subsystem from


the SCU maintenance panel.

In addition, the 3672 SCU performs the following


miscellaneous functions'

4. Miscellaneous Errors

Overflow Record

a. Seek Errors - If a Seek error is detected by the


SCU. the Seek is retried before it is considered a
malfunction.

The SCU can operate on records that extend past the


end of a track and continue on the next track.
Overflow records are indicated as such in the flag
byte, and when writing such records, the Write
Special Count Key. and Data command must be used.

b. Defective Track If a defective track is


detected, the channel is notified and the user's
control program may flag the defective track and
reissue the command to cause the data to be
written on an alternate track.

Multitrack Operation
On Read or Search commands the operation will
become a multitrack operation if bit 0 of the respective
command is a 1. This allows an entire cylinder to be
searched with one command.

Channel Disconnect Operation


When operating in conjunction with a block multiplex
channel. the SCU can disconnect from the channel
when a command requiring mechanical motion is
being executed. This disconnect can occur even
though chaining is indicated by the channel. The SCU
retains all information required to control a disconnected CCW chain for each attached disc spindle.

c. Command Overrun - If during command chaining the channel fails to meet the real-time
chaining constraint of the Subsystem. the SCU
will initiate a retry of the failing command.
If during data transfer the
d. Data Overrun channel fails to meet the real-time requirements of the Subsystem. the SCU will
initiate a retry of the failing command.

The SCU contains counters which enable it to maintain a statistical log of the usage and error occurrences for each drive in the Subsystem. The usage
counters count the total number of access motions
and the total number of data bytes transferred to the
channel. The error counter is an accumulation of the
total number of correctable or uncorrectable Read
errors and Seek errors. The counter data is periodically transferred to the channel for system logging.

Error Correction
The error correction function provided in the 3672based Subsystem allows the SCU to detect errors and
correct these errors through the use of Command
Retry. The correction of a failing command through
Command Retry is used for the following errors:
1. If the data is determined to be uncorrectable (i.e.,
the error exceeds the correction code capability),
the channel is signalled and the command may be
retried under control of the channel.
2. If the data error is correctable and it occurs in the
data fiele;! of a record, the SCU passes on to the
channel the information necessary to correct the
data.
3. If the data error is correctable and occurs in the
Home Address, Count or Key fields. the SCU
corrects the data which is being held in a buffer

Usage and Error Logging

also can be run from the Field Engineer (FE) panel on


the SCU. Error messages are displayed on the same
FE panel.
The control of the 3673 Controller and 367X Disc Drives,
and the data processing to provide the recording format
on the disc track. is defined by the microprogram
routines which are resident in the WCS portion of the
Microprocessor. Logic signals to operate the disc drives
and sequence the CPU channel are obtained from latches
set by the microprogram in the Controller Interface and
Channel Interface. respectively. The latches for a given
function are grouped into a register, and binary data is
loaded into these registers in timed sequences to
stimulate the associated hardware. The microprogram
can interrogate the status of the different groups of
hardware by testing the state of the appropriate latches.
These latches are also grouped by function into registers.
Testing is accomplished by selecting the desired register
and isolating the bit or bits to be investigated. The
microprogram sequence is then alternated by branching
if the state of the latch(es) corresponds to the condition
for branching.
There is a prescribed interaction involving stimulus and
response in which the microprogram controls both the
timing and the sequence. The loading and accessing of
registers is defined by the operand fields of the
microinstruction word and the register's data is moved
over internal logic buses.

1.7.2 3673 Disc Drive Controller


The 3673 Disc Drive Controller functions as an electronic
interface unit that provides drive-related control functions to the DDMs. It accomplishes this by decoding and
executing tag instructions from the Storage Control Unit.
These tag instructions are divided into three groups:
drive-related
tags,
controller-related
tags.
and
drive/controller-related tags.

Diagnostics
The control unit contains microprograms which are
used for online, inline, and offline servicing of the
SCU, Controllers. and DDMs. When the FE inserts the
service module plug into one of the drives. the drive
becomes offline to the operating system and online to
the control unit. Using the proper Memorex diagnostic
programs, diagnostics can then be run and error
messages displayed via the USing system. Diagnostics

Drive-related tags are received over the CTL-I interface of


the Controller and passed directly to the drive. The
Controller does not operate on this group of tags and is
completely transparent to their occurrence. except for
certain data checking validity functions.

Controller-related tags are operated on by the Controller


These tags are not passed to the drive.

Controller/drive-related tags are operated on jointly by


the Controller and drive. The read/write tags fall into this
group as well as certain diagnostic tags used for
diagnostic purposes
Functionally. the Controller is divided into three logical
areas as shown in Figure 1-3: the Controller Interface
logic. the Drive Interface logic, and the Read/Write logic.
The interface groups provide Signals between the Controller and the SCU and drives by means of line drivers
and receivers. The Read/Write logiC consists of two
subgroups: the read/write data transfer and read/write
control. The data transfer group consists of registers.
multiplexers. and other logic required to move the data
through the Controller to either the SCU. if reading data,
or to the selected drive. if writing data. This group also
contains the serializer /deserializer (serdes) used to
assemble data into either serial form for writing onto the
disc or into parallel form (deserialize) if reading from the
disc. The control logic consists of latches. counters. and
related logic that supervise the transfer of data under
control of tag instructions.

1.7.3 3670 (3675) Disc Drive Module


The 3670 (3675) DDM consists of two independent disc
storage drives. Each drive (spindle) stores and retrieves
data on a MEMOREX Mark X. IBM 3336-1. or equivalent
disc pack (MEMOREX Mark XI, IBM 3336-11. or equivalent
for 3675 DDMs). Disc pack information written on one
3670 (3675) Drive may be retrieved or updated on any
other 3670 (3675) Drive. IBM 3330 (3330-1). or
equivalent.
Each drive is capable of seeking to anyone of 404 (808)
cylinders plus 7 alternates. selecting anyone of 19
heads. and transferring information to the control unit at
a nominal data rate of 806 kilobytes per second.
Up to four modules can be attached to the 3673
Controller. The primary functions performed by each of
the disc drives are as follows:

Position the access mechanism to a cylinder.

Select a head.

Read or write data.

Respond to commands given by the Storage Control


Unit.

1-4

L.. READ DATA

......
FLEXIBLE DISC

ADDRESS BUS
MEMORY OUT BuS
~

~~

MEMORY OuT BUS ..

CONTROLS ..

.. FD ADDRESS BUS
~

.~

~~
BRANCH

FLEX DISC

INTERFACE

CONTROL STORAGE
CONTROL CI RCUITS

.i~

FLEX
DISC
DRIVE
CONTROLS

I
I
I
I

D
BUS

REGISTER

SR
REGISTER

I
I
I

"'"

.4 ~

.. WRITE BUS

CONTROL STORAGE
(MEMORY)

INCR
REGISTfR
~,

~r

___ FDF READ DATA

TOA/F
MUl TlPLEXER

FE PANEL

'~D2

f-----

....

.......

FE CONTROLS

-..

.. SW!.',
~

MICROPROGRAM DATA
FLOW CONTROLS

DI. DO. CO. Rwe REGISTER BITS

f - -...

---

-- -- ------

..

CTl CONTROLS

TO BRANCH
CONTROLS

4~

.. ~

DATA FROM
BO. CHF CI
BRANCH
CONDITIONS

....

CONTROL" ....

--..

CONTROLLER
INTERFACE

I
CTl DO BUS

CONTROLLER/DRIVE
INTERFACE
AND CONTROLS

III

....

BRANCH
CONDIT,O'<S

I~
I'
READ/WRITE
(SERDES)

.4~
READ

I
STORAGE CONTROL UNIT

HEAD
POSITIONING

po

SERVO CONTROL ...

01 BUS

3673

.~

CONTROLLER

SERVO
PULSES

SERVO PULSES

I
I
I

I
3672

...

ORDER
DECODE

I
1--- - - - - -

I
I
I

CHANNEL
INTERFACE

1~

...

BUS :-

DRIVE DO BUS

~ DRIVE

DATA TO
BI. CT. CHC
REGISTERS

--

..

HEAD ADDRESS ...

~RIVE OR~ER

.
..

... CTl 01 BUS

... INDEX
r""

_..J

DRIVE CONTROLS

cr L ORDER BUS --.

I
TO BRANCH
CONTROLS

I
I

REGISTE R ..

R/W
CONTROL

~Ir
WRITE DATA ..

..,..

WRITE

I
I

HEAD
MOTION

~r

......

DISC
PACK

READ/WRITE
CONTROLS
.... READ DATA

.....
3670/3675

DISC DRIVE

Figure 1-3. Disc Storage Subsystem Operational Block Diagram

1-5

The drive assembly consists of a motor and a spindle


drive. The removable Mark X (Mark XI) pack is mounted
on the spindle, which is driven at 3600 RPM. The motor
cannot be started unless the pack access cover is closed.
The access cover is locked automatically whenever the
spindle motor is energized or a stop sequence is in
progress.
The access mechanism consists of an electromagnetic
voice coil motor and a carriage assembly on which the
recording heads are mounted. Major assemblies of the
access mechanism are:

Actuator
An electromagnetic actuator moves the carriage
assembly to any of 411 (815) cylinder positions.

Carriage Assembly
The carriage assembly consists of 1 9 Read/Write
heads and one servo head mounted on a carriage.

Access to any cylinder is achieved by a linear voice coil


motor which drives a carriage on which the data heads
and servo head are attached. The servo head is used to
position and lock the carriage on the desired cylinder by
servoing on information which was prerecorded on the
disc pack at the factory. Once the desired cylinder has
been reached and the servo system has locked on the
servo track, any data head can be electronically selected
and the data is recorded on, or retrieved from, the track.

1.8 DATA INTERFACE REQUIREMENTS


All interface cabling is provided with the 3672-based
Subsystem. The channel interface circuitry in the Subsystem is compatible with the liD interface requirements
established by IBM.
The Memorex FE can establish priority to any 3672based Subsystem. This capability enables users of multiple 3672-based Subsystems to reserve individual
systems for selected applications. For example, highactivity applications can be reserved for a system assigned a normal priority, while low-activity applications can
be reserved for a system assigned a low priority.

1-6

SECTION 2. COMMANDS
2.1 GENERAL
This section defines and discusses the commands per
formed by the 3672 SCU during execution of liD
instructions. The section is divided roughly into two
halves. The first half contains instruction and command
word formats, followed by a detailed description of each
command in both narrative and flow diagram form. The
second half discusses command-related hardware
operations, such as command retry, rotational position
sensing, and overflow records.

The status of the CPU existing at the time of the change


is stored In the Program Status Word (PSW). After the
specified instruction has been executed, the CPU can
return to the problem state and continue the interrupted
program by reloading the PSW originally stored when the
program entered the supervisor state. In the supervisor
state, the CPU can execute the following 1/0 instructions: Start 110, Start I/O Fast Release, Test I/O, Halt
110. and Halt Device.

The I/O Instruction format

IS

14 15

as follows

16

(NOT USED)

OP

19

20

B,

D,

0-7

START 1/0 FAST RELEASE

8-14

Not Used.

15

Set to 1 for Start I/O Fast Release and Halt


Device instructions.

Start liD
Start liD Fast Release
Test liD
Halt liD
Halt Device
Channel Address Word
Channel Command Word
Channel Status Word
Program Status Word

2.3.2 1/0 Instructions


The I 0 Instructions In the CPU program initiate liD
operations of the 3672-based subsystem. The operations
are controlled by the channel using commands read from
main storage. Arithmetic and logical (decision) operations
are performed while the CPU is in the problem state: for
liD operations, the CPU must be In the supervisor state
The CPU is changed from the problem to the supervisor
state when a Supervisor Call Routine is executed or
when an I/O interrupt sequence occurs.

Must be zero.

21-23

CHNL ADDR (Channel Address)

24-27

START 1/0
The Start I/O instruction initiates an I/O operation upon
detection that the addressed channel, SCU, and drive are
available.

The follOWing material descnbes the formats for the


following liD Instructions, channel words, and the
Program Status Word:

16-20

Designates channel address.

Table 2-1 summarizes the command groupings and


coding information for the 3672-based subsystem commands. Detailed descriptions of individual commands are
given on following pages.

2.3.1 Introduction

Not Used

31

2.2 COMMAND SUMMARY

2.3 INSTRUCTION AND WORD FORMAT

0-1 5

SCU ADDR (SCU Address)

OP (Operation Code)

Designates SCU address.

Specifies the operation to be performed.


29-31

The Start I/O Fast Release instruction initiates an liD


operation upon detection that the addressed channel is
available. The SCU and dnve are assumed to be already
available. If not. an liD interrupt sequence occurs to
indicate an unavailable condition.

16-19

The Test 110 instruction sets the condition code in the


PSW to indicate the status of the addressed channel,
subchannel, SCU, and disc drive.

Designates drive address.

B, (Base Address Register Location)


Specifies the address of a general register in
the CPU. The register is 32 bits long, but only
the low-order 24 bits are used.

TEST 1/0

20-31

0, (Displacement)
An immediate field added to the contents of the
register at B, to develop bits 16-31 of a 32-bit
result. This result identifies the channel and
device addressed by the instruction and has the
following format

Halt liD
The Halt I/O instruction terminates the operation in
progress at the channel

HALT DEVICE
0

The Halt Device Instruction terminates the operation in


progress at the SCU without interfering with other 1/0
operations at the channel. This instruction is used
instead of Halt I/O to terminate an operation on a device
attached to multiplexer channels.

15
(NOT
USED)

16

DEV ADDR (Device Address)

20

00000

21

23

CHNL
ADDR

24

27

SCU
ADDR

28

31

DEV
ADDR

2.3.3 Channel Address Word


The Channel Address Word (CAW) is read by the channel
from main storage location 72 when a Start 110 or Start
I/O Fast Release instruction is issued. Bits 0-3 of the
CAW form the protection key for all commands
associated with the liD instruction. The protection key
establishes the privilege of access to the particular main
storage locations (for example, whether data can be
stored or read). The command address in bits 8-31
designates the address of the first Channel Command
Word (CCW). The three low-order bits of the command
address must be zero to specify the CCW on double-word
boundaries. CAWs are read by the channel hardware.
The information must be set up in main storage location
72 prior to issuing the I/O instruction.
The Channel Address Word format is as follows

3
KEY

0000

31

8
COMMAND ADDRESS

2-1

TABLE 2-1

COMMAND SUMMARY

HEX CODE
COMMAND GROUPINGS
CONTROL

COMMAND NAME

CONTROL COMMANDS start operations not involving a transfer of data records


between the SCU microprocessor and system main storage. These bytes enable the
operation to take place and are parity checked during transfer. For most control functions. the entire operation to be started is specified by the command code. If not so
specified. the additional information needed is fetched from main storage.

I
SENSE

READ

WRITE

SEARCH

SENSE COMMANDS determine the status of the Subsystem and identify the specific
nature of any errors or unusual conditions that have occurred.

WRITE COMMANDS transfer information from the system main storage to the
SCU for recording on a disc pack. While
writing a record on a disc track, the
Controller appends the appropriate correction code bytes to each record area.

Update Write Commands (nonformatting) operate on previously formatted tracks to update existing
records.
Formatting Write Commands initialize
tracks and records, and establish the
length of the areas within each
record.

SEARCH COMMANDS transfer a specific number of bytes from system main storage to
the SCU. While executing a command, the channel operates in the Write mode while the
_disc storage operates in the Read mode. Incoming data from main storage is compared
'With outgoing data from disc storage. When search criteria specified in the command are
satisfied. the status modifier bit is set. This bit causes the channel to skip the next CCW
in the chain and fetch the next command from main storage. Each search command
operates on one record at a time. To search another record, the command must
be reissued.

tSet RPS command is valid only if the 2860 Attachment Feature is installed. See paragraph 3.12.

3672.21-0001-10/75

NO-OPeration
SEEK
SEEK CYLINDER
SEEK HEAD
SPACE COUNT
RECALIBRATE
RESTORE
SET FILE MASK
SET SECTOR
DIAGNOSTIC LOAD
DIAGNOSTIC WRITE
SET RPS t

03
07
OB
lB
OF
13
17
lF
23
53
73
2F

SENSE I/O
READ AND RESET BUFFERED LOG
READ DIAGNOSTIC STATUS 1
DEVICE RESERVE
DEVICE RELEASE

04
A4
44
84
94

READ
READ
READ
READ
READ
READ
READ
READ

READ COMMANDS transfer information from a disc drive to the system CPU. All
except Read IPL and Read Sector may operate on overflow records and in multitrack
as well as singletrack mode. On all read commands. the SCU examines correction code
bytes to check 'the validity of each record area, and adds a parity bit to each byte.

--i{

SINGLE MULTITRACK TRACK

DATA
KEY AND DATA
COUNT. KEY. AND DATA
RO
COUNT
HOME ADDRESS
IPL
SECTOR

WRITE DATA

06

SPECIAL COUNT, KEY, AND DATA

HOME ADDRESS EQUAL


10 EQUAL
10 HIGH
ID EQUAL OR HIGH
KEY EQUAL
KEY HIGH
KEY EQUAL OR HIGH

BIN (PACK)

86
BE
9E
96
92
9A

-B9
Bl
01

F1
A9
C9
E9

----..

I
I

00
0OOO-019A

.-3670

0OOO-032E

.-3675
00 100-12

HEX CODES OF PERMIT COMMANDS


WRITE
SEEK
4

05

001 10

01 1 15 1 19 1 11

vlv

vi

V
V

\lJ

07 1 OB 11B

1-../

NONE PERMITTED

vlv

vlvlv Iv
V I V Iv
I V Iv
Iv
I
NONE PERMITTED

HEX CODE OF PREREQUISITE COMMAND


(ANY ONE)
15

19

29/A9*

~V

0
00

HEAD

HEX CODES OF
DRIVE ADDRESS BYTES

CYLINDER

MASK
BYTE BITS

39
31
51
71
29
49
69

~ WRITE

22

SEARCH
SEARCH
SEARCH
SEARCH
SEARCH
SEARCH
SEARCH

WRITE COUNT, KEY, AND DATA

...

WRITE RO
WRITE HOME ADDRESS
ERASE

WRITE KEY AND DATA

...

OE
lE
16
12
1A
02

05
00
10
01
15
19
11

SUMMARY INFORMATION

31/Bl*

39/B9*

V
V
vi

vi
V

10

V
NOT REQUIRED

STATUS
BYTE
1

BITS SET IF
SEARCH CRITERIA ARE SATISFIED
CSW
NAME
BIT
33

STATUS MODIFIER

36

CHANNEL END

37

DEVICE END

* Search must not be truncated.

2-2

0-3

0-7

KEY (Protection Key)

4-7

main storage. When this bit is 0, normal


transfer of data takes place.

CMD CODE (Command Code)


Specifies the operation to be performed. Either
the two or four low-order bits of the command
code identify the type of operation to the
channel as follows:

Identifies the storage protection key for all


commands associated with Start 110. This key
must match the storage key.

8-31

Command Address
Designates the location of the first CCW in
main storage.

When set to 1, causes the channel to generate


an interrupt condition upon reading the CCW.
When this bit is 0, normal operation takes
place.

XXXX XXll
XXXX Xl00
XXXX XXl0
XXXX XXOl
XXXX XOOl
37-39

For every CCW other than one specifying TIC


(Transfer in Channel), these bits contain zeros.
Violation of this restriction generates a
program-check condition.

The channel distinguishes Write, Control, Read,


Sense, or Transfer-in-Channel (TIC) operations.
All eight bits are transferred to the SCU when
1/0 operations are initiated.

2.3.4 Channel Command Word


8-31
The Channel Command Word (CCW) is read by the
channel from the address specified in the CAW. The first
CCW specifies the operation to be performed, the main
storage locations to be used; and the action to be taken
when the operation is completed. If available when it
receives the CCW, the channel attempts to select the 1/0
device specified in the 1/0 instruction by sending the
address to all attached SCUs. If the addressed 1/0 device
is attached to the channel and has power on, the
command code portion of the CCW is sent to the SCU,
which responds with an initial status byte to the channel.
At this point. the Start 110
instruction is finished,
releasing the CPU to perform the next instruction.

The results of the attempt to initiate execution of the


command are indicated by the condition code in the
Program Status Word (PSW). If the 1/0 operation was not
started, new status information containing the reason for
this conditioI' is normally set In the Channel Status Word
(CSW) The CCW format IS as follows

40-47

Not Used.

48-63

Count
Specifies the number of eight-bit byte locations
in main storage area designated by the data
address.

CD (Chain Data)
When set to 1, specifies chaining of data.

33

8-31

Command Address
An address that is eight positions higher than
the address of the last CCW used.

32

Attention-not used.

33

Status Modifier
Set for three conditions (1) Whenever a Search
IDIKey High, Search IDIKey Equal, or Search
IDIKey Equal or High command has been
executed and the search criteria satisfied; (2)
With Busy bit to indicate SCU Busy, or (3) With
Unit Check and Channel End to indicate retry
status.

Data Address
Specifies the location of a two-byte address in
main storage. This is the address of the area
associated with data transfer operations.

32

Must be zero.

PCI (Program Control Interruption)

36

Must be zero.
Control
Sense
Read
Write
Search

4-7

CC (Chain Command Flag)


When set to 1, and when the CD flag is zero,
specifies chaining of commands. It causes the
operation specified by the command code in the
next CCW to be initiated upon normal completion of the current operation.

34

Set if an SCU Busy status has been generated


previously and the busy condition has been
terminated. SCU End may be given with any
device address recognized by the SCU.

2.3.5 Channel Status Word


The Channel Status Word (CSW) informs the program of
1/0 device status or the conditions under which an 1/0
operation was terminated. It is stored at main storage
location 64. The CSW is formed or changed during 1/0
interruptions and instruction execution. Status stored in
the CSW remains unchanged until a subsequent interrupt occurs or a new 1/0 instruction is processed.

35

34

31

32

39 40

47

48

Set at the end of each channel command.


KEY

0000

COMMAND
ADDRESS

DEV
STATUS

CHNL
STATUS

COUNT

37

CMD
CODE

31

32

33

34

35

DATA
ADDRESS

CD

CC

SLI

SKIP
FLAG

FLAGS

36 3739 40 47 48 63
PCI

000

(NOT
COUNT
USED)

35

Device End
Indicates that an access mechanism is free to
be used.

The Channel Status Word format is as follows:


7

Channel End

63

SLI (Suppress Length Indication)


When set to 1, suppresses an incorrect length
condition except when the CCW count is not
exhausted, in which case Channel End is
present and data chaining is indicated.

Busy
Indicates that the selected drive is busy. In
conjunction with bit 33, indicates the SCU is
busy.

36
0

SCU End

Skip Flag
When set to 1, specifies suppression of a
transfer of information to main storage during a
Read or Sense operation. Checking takes place
as though the information has been placed in

0-3

KEY (Protection Key)


The storage protection key used in the chain of
operations.

38

Unit Check
Set whenever an unusual or error condition is
detected.

2-3

39

40-47

Unit Exception

2.3.5.3 ENDING STATUS

2.3.5.5 CONTINGENT CONNECTION

Indicates an End-of-File has been detected


during a Read RD. Read IPL. Read Count. Key.
and Data. Read Key and Data. Read Data. Write
Key and Data. or Write Data operation. It
results from a data length of zero being
detected in the count area of a record. When
detected. no data is transferred from the data
area. If the key length is not zero. the key area
is transferred.

In most cases. Channel End and Device End are


presented as the normal ending sequence for an opera
tion. The exceptions are noted in the individual command
descriptions which follow. If an error occurred during the
operation. Unit Check will accompany the Channel End
and Device End status.

A contingent connection is established in the SCU after


the channel accepts a status byte containing Unit Check.
The connection lasts until one of the following conditions
occur:

CHNL STATUS (Channel Status)

A pending status condition may exist for the SCU or disc


drive. Status is pending for the SCU if one of the
following conditions has ocurred:

Indicate channel condition as follows:


Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

48-63

40:
41:
42:
43.
44:
45:
46:
47'

Program-Controlled Interruption
Incorrect Length
Program Check
Protection Check
Channel Data Check
Channel Control Check
Interface Control Check
Chaining Check

Count
The residual byte count from the last CCW
used.

0-7

A command (other than Test 110 or No-Op)


receives an initial status byte of zero for the
SCU and disc drive address which generated the
Unit Check.

A Selective or System Reset occurs.

A disconnect was Signalled after a command


was issued. but before Channel End was
accepted.

Busy. Channel End. or Unit Check status was


stacked by the channel.

Zero status was stacked by the channel


response to a Test 110.

SCU Busy was presented to the channel

Unit Check was detected for an operation after


Device End had been cleared.

Device End status for a Set Sector command


was stacked by the channel.

The Program Status Word (PSW) contains the status of


the using system. Two PSWs are associated with subsystem interrupt conditions: an old PSW which contains
the status information of the using system existing at the
time of the interrupt, and a new PSW which is used to
control instruction sequencing and hold the status of the
using system relative to the program being executed. By
storing the new PSW during an interruption. the CPU
status is preserved for subsequent inspection by the
program.

Status pending for the SCU causes it to appear busy for


all devices except the device for which the status
condition exists. Unless the SCU is busy. it will request
service to clear the pending status condition. Status is
cleared when presented to. and accepted by. the
channel. Status is pending for a drive if one of the
following conditions has occurred:

Loading a new PSW causes the state of the CPU to be


initialized or changed to branch to a new instruction
sequence. If. at the conclusion of an interrupt routine. an
instruction is executed which restores the old PSW as
the new PSW. the using system is restored to the state
existing prior to the interruption. and the interrupted
routine continues.

In

2.3.6 Program Status Word

2.3.5.1 STATUS PRESENTATION


Status is presented twice (initial status and ending
status) for all commands except those commands that
require access motion. and immediate commands not
chained from write commands. Seek and Seek Cylinder
commands present status three times: initial status.
Channel End status (after transfer of the Seek Address).
and Device End status (after the access IS positioned).

2.3.5.2 INITIAL STATUS


The initial status byte is zero for Test 110 and all nonimmediate commands unless one or more of the following conditions exist: SCU is busy. a status condition is
pending. a Unit Check occurred. or initial status indicated
Command Retry. Immediate commands (for example.
commands not requiring data transfer) present Channel
End and Device End In initial status.

Designates the system mask as follows:


Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

2.3.5.4 PENDING STATUS

During the contingent connection state. the SCU appears


busy to all SCU and device addresses other than the
address for which the contingent connection was established.

Channel End appears alone.

Busy status is presented.

The drive has gone from not-ready to ready.

System Mask

8-11

0: Channel 0 Mask
1: Channel 1 Mask
2: Channel 2 Mask
3: Channel 3 Mask
4: Channel 4 Mask
5: Channel 5 Mask
6: Channel 6 Mask
7: External Mask

KEY (Protection Key)


Identifies the storage protection key.

12

Must be zero.

13

M (Machine Check Mask)

14

W (Wait State)

15

P (Problem State)

16-31

INTRPT CODE (Interruption Code)

32-33

ILC (Instruction Length Code)

34-35

CC (Condition Code)

36-39

PROG MASK (Program Mask)

The PSW format is as follows:


Designates the Program Mask as follows:

Status pending for a disc drive causes the SCU to request


service when both the SCU and disc drive are not busy.
The status is cleared when presented to. and accepted
by. the channel.

SYSTEM
MASK

811 121315 16

KEY

M.
W.
P.

31 ~233 3435 36 39 40

INTRPT
CODE

ILC

CC

Bit
Bit
Bit
Bit

63

PROG INSTRUCTION
ADDRESS
MASt<

40-63

36:
37:
38:
39:

Fixed-Point Overflow Mask


Decimal Overflow Mask
Exponent Underflow Mask
Significance Mask

Instruction Address

2-4

2.4 CONTROL COMMANDS

Controls provide tlmmg for data transfer and check


parity of data transferred.

c Arithmetic Logic Unit which calculates the


difference between present head location and new
address in the stored command code.

2.4.1 Operation Block Diagram


The block diagram shown in Figure 2-1 and the text
below provide an Illustration of the System Control
operation during the execution of a typical control
command-that of a Seek command.

2.4.1.1 SYSTEM
The CPU executes a Start liD instruction. Channel
Address Word (CAW) specifies the main storage location
of the first Channel Command Word (CCW). The SCU and
drive to be used are specified by the Start liD
instruction

2.4.1.2 SYSTEM INTERFACE

Channel executes the CCW to transfer the Seek


command and Seek address to the SCU

Channel disconnects until the SCU signals with


Device End status

Controller Interface provides for communication of


control, status. and address information between the
SCU, Controller, and the Disc Drive Module.

2.4.1.3 STORAGE CONTROL UNIT

Channel Interface provides for communication


between the SCU microprogram and the System
Interface.

Microprocessor contains the subsystem memory and


all hardware for control of data flow between the
Channel Interface and Disc Drive Module. Components are:

SYSTEM
(CENTRAL
PROCESSING
UNIT)

..

SYSTEM
INTERFACE
ICHNL AND
CTRLS)

2.4.1.4 DRIVE CONTROLLER


The Drive Controller provides communication path for
control. address, and status information between the
SCU and the Disc Drive Module.

2.4.1.5 DISC DRIVE MODULE

b. Registers which decode and store the command


code, and receive record-length control information

INTERFACE

L - -

- -

!....--!

SE~

Head Positioning Controls determine the direction and


amount of movement of the Head Access Mechanism.

f,lPROG

r--MODULE STATUS
It.FORMATION

CONTROLLER

~c;-- ~I
I

I --1
-

PROCESSOR

--

~A~

ZON-:;:-RDLt7R 1
1
INTERFACE...J

L.. -

the

head-arm

Disc Pack provides signal for each track passed from


the servo surface in the pack.

Each control command is described in the following


pages by means of a narrative description and a flow
diagram. The number(s) adjacent to each flow diagram
block refer to a routine in the 3672 ~Program Flowchart
Manual. Numbers not in parentheses refer to mainline
routines, while numbers within parentheses designate
'subroutines called by the mainline routines.
Many of the flow diagrams begin with an INITIAL
SELECTION sequence. The flow diagram for this sequence is shown in Figure 2-1A.

DISC DRIVE MODULE

..

BUS OUT
.. BUS IN

TAG OUT

DATA

---.,

IL _
INTERIOR
FNOT~~~~~:ANEL
___
_ _ _ _ _IFEI
_

Head Access
assembly

2.4.2 Command Descriptions

Control
Storage that contains
a Writable
microprogram routines that define the data handling and drive control.

r ~H:N~L- ~

Mechanism l moves

FE Interface provides controls and indicators for


facility operation (at the Operator Panel) and
maintenance (at the Interior Control Panel).

STORAGE CONTROL UNIT


SEEK CONTROL
INFORMATION

They maintam head position. and signal to the SCU


when the desired new location is reached.

...JI

TAG IN

.-

370

Figure 2-1 Seek Operation Block Diagram

3672 .21 -0001-10/75

2-5

302
ALLOW CHANNEL
DISABLES IF
NO OUTSTANDING
DRIVE INTS
FOR CHANNELS

YES

302
SCAN BUFFER
FOR ANY
STATIC
INTERRUPTS

CALCULATE
ADDRESS OF
POLLED
DEVICE

302

UPDATE DE
READIES IF
REQUIRED

SET
SUPPRESSIBLE
REQUEST IN IF
ANY STATIC
INTERRUPTS
NO

RAISE
REOUEST
IN

302

302
"OR" CU END
IN ENDING
STATUS IF
REOUIRED

302

POLL DRIVES
FOR
INTERRUPTS

SET CUE
INDICATOR

PRESENT
CU BUSY
STATUS

302

"OR" CU END
IN ENDING
STATUS IF
REQUIRED

303

SET
NONSUPPRESSIBLE
REQ IN IF ANY
DYNAMIC
INTERRUPTS

FETCH
DDW 0 AND 1

NO
YES

303

YES

RESET
INDEX
YES

303
SET UP
STRING POINTER
IN
ROI7

080

Figure 2-1 A. Initial Selection Sequence, Block Diagram (1 of 6)

3672.21-0001-10175

2-SA

303
303
FETCH
CHANNEL
INTERRUPT
WORD

SET JlPROG
DETECTED
ERROR

SET NOT
SINGLE STRING
SELECTION,
/1PROG DETECTED
ERROR

RESET
INDEX

SET BUSY
IN DRIVE
STATUS

CLEAR
DRIVE
STATUS

303

OBTAIN STATUS
FROM DRIVE
AND STORE IN

S~LECT

DEVICE

G7

303
INITIALIZE
CMD OUT TIMER.
RAISE OP IN
AND ADRS IN

SET IlPROG
DETECTED
FRROR

PRUPAGAT~

SELECT
OUT

SET "LONG
BUSY ON
r:HNL INIT
INDICATOR

Figure 21 A. Initial Selection Sequence, Block Diagram (2 of 6)

3672.21-0001-10/75

2-58

CLEAR
CHANNEL
COMMAND

00 NOT
SET NEW
COMMAND
IN R3/B

FETCH
SCB

303
YES

RESET RETRY
IN PROGRESS
AND DEF/ALT

NO

SET
LOCK
CHANNEL
SWITCH
MAKE RESERVE
OR RELEASE
CMOS LOOK
LIKE SENSE
CMD

YES

303

303

SET DE
STATUS

SET UC
STATUS

Figure 2 -, A Initial Selection Sequence, Block Diagram (3 of 6)

3672.21-0001-10/75

2-5C

V
L'"

303
SET
FORMAT 2,
MESSAGE 0

SET LONG
CONNECT
FOR THIS
INTERFACE
FETCH USAGE
COUNTER
FOR THIS
STRING

303
READ 3/6
CODE
FROM DRIVE
YES

303

READ DEVICE
TYPE AND
CAR/CHAR

ENABLE
CHECK-2

303
SET
FORMAT 6,
MESSAGE 0

303
SET
FORMAT 2,
MESSAGE 0

FETCH ERROR
LOG BLOCK

303
SET
INTERVENTION
REQUIRED

SET
ENVIRONMENTAL
DATA
PRESENT

Figure 2-1 A. Initial Selection Sequence, Block Diagram (4 of 6)

3672.21-0001-10175

2-5D

YES

YES

303
YES

SET
PRIME
DEVICE
INTERRUPT

SET
EQUIPMENT
CHECK,
PERM ERROR

303
SET
STACK AT DEVICE
AND
UNTtMED

303

303
SET
FORMAT 1,
MESSAGE 5

SET
SIP

303
SET
EQUIP CHECK,
WRONG CCW

NO

303
RESET
POLL ENABLES
ON OTHER
CHANNELS

YES

YES

307

304

306

11

Figure 2-1 A. Initial Selection Sequence, Block Diagram (5 of 6)

3672.21-0001-10/75

2-5E

303

303

SEl
CHAINING
AND
UNTIMED

SET
DE
OWED

YES

YES

303

303

INITIALIZE
CMD OUT TIMER.
RAISE OP IN
AND ADRS IN

RESET
RETRY
PARAMETERS

YES

'OR" BUSY
INTO
STATUS
SET
BUSY

SET
ASSIGN AND
RESET PE'S
ON OTHER CHNLS

303
SET
DEVICl
BUSY
STATUS
DROP
ADRS IN

12

Figure 2-1 A. Initial Selection Sequence, Block Diagram (6 of 6)

3672.21-0001-10 75

2-5F

NO-OP
Command Code 03 (hex)

(Figure 2-2)

An immediate command which causes no action at the


addressed drive.

INITIAL STATUS-Channel
presented

End

and

Device

End

SPECIAL REQUIREMENTS-Indiscriminate usage must


be avoided, since a No-Op resets orientation information
causing all or part of record to be skipped. For example:

SEEK
Command Code 07 (hex)

(Figure 2-3)

Transfers the six-byte seek address shown below to the


SCU.
ADDRESSB'fT~S

DATA ADDRESS-Not checked for validity, but should


not exceed addressing capability.
FLAGS-SLI flag must be set to avoid an incorrect length
indication.
COUNT-Must not be zero. A zero count sets the
Program Check bit (bit 42) in the CSW.

A No-Op inserted between Read Count and Read


DClta causes data of the following record to be
skipped.
A No-Op inserted between a command which
reads the data field of record n - 1 and a command
which must process the Count field of record n
may skip record n and process the Count field of
record n + 1.

(HEX)

CYLINDER

BIN

HEAD

BYTE
0

BYTE
1

BYTE
2

BYTE
3

BYTE
4

BYTE

3670

00
00

00
00

00
01

00 TO FF
00 TO 9A

00
00

00 TO 12
00 TO 12

3675

00
00
00
00

00
00
00
00

00
01
02
03

00
00
00
00

TO
TO
TO
TO

00
00
00
00

00
00
00
00

FF
FF
FF
2E

TO
TO
TO
TO

12
12
12
12

The SCU selects the drive, moves access to the proper


cylinder, and selects the proper head. Any access motion
required is initiated after the seek address has been
transferred. Limitations on address byte values are:

Channel End is presented after the seek address has


been transferred. Device End is presented with Channel
End if no movement is required, or after access is
positioned if movement is required.
DATA ADDRESS-Specifies main storage location of the
seek address. Checked for both validity and correct parity
by the SCU. If an invalid seek address or parity error is
noted, the command is not executed; Unit Check,
Channel End, and Device End are presented in ending
status; and a subsequent sense command indicates
either command reject (invalid condition) or bus-out
parity error (parity error).
FLAGS-Can be used at discretion of the programmer.
COUNT-Must be six or greater. If less than six, command is not executed; Unit Check, Channel End, and
Device End are presented in ending status; and a
subsequent sense command indicates command reject.
INITIAL STATUS-Normally zero.

Bytes 0, 1, and 4 must be zero.

Bytes 2 and 3 together must not exceed 410


(dec) for 3670. or 814 (dec) for 3675.

Command execution does not require preceding CCW.

Byte 5 must not exceed 18 (dec).

File mask must be set to allow seeks.

SPECIAL REQUIREMENTS-

2-6

This command requires the Seek Address to be transferred to the SCU. The
SCU selects the Controller and drive
and transfers Seek Control information
to the drive. The drive controls the
movement of the access to the new
position.

SEEK CYL
Seek Cylinder command is the same
as Seek command.

Tile No-Operation Command causes no


action at addressed device. This command causes the SCU to lose orientation to the track format.

SEEK HEAD
Seek Head is the same as the Seek
command except only the head address is changed.

SEEK ADDRESS
RST TRK
ORIENTATION
05F
306

BIB

C I C

3670

ZEROS

0-410

ZERO

0-lB

I SCU

3675

ZEROS

0-814

ZERO

0-18

II

r--

- -*,

RECONNECTS
TO CHANNEL
(IF BLOCK MUX
CHANNEL)

L ____ ...I

PRESENT
CHNL END AND
DEVICE END
TO CHNL

306
SCU SENDS
'DEVICE END
.TO CHNL
DRV SENDS
SEEK COMPLETE
TO SCU

306

306
. RST SEEK
COMPLETE
INTERRUPT

306 if not Block Multiplexer


302. 303 if Block Multiplexer

If 2860 Attachment Feature installed.


See paragraph 3.12.

Figure 2-2. No-Op Flow Diagram

3672.21-0001-10175

It
I
I

307,302

Figure 2-3. Seek Flow Diagram

2-7

SEEK CYLINDER
Command Code 08 (hex)

(Figure 2-3)

Transfers the six-byte seek address shown below to the


SCU.
ADDRESS BYTES (HEX)
CYLINDER

BIN

HEAD

BYTE
0

BYTE
1

BYTE
2

BYTE
3

BYTE
4

BYTE
5

3670

00
00

00
00

00
01

00 TO FF
00 TO 9A

00
00

00 TO 12
00 TO 12

3675

00
00
00
00

00
00
00
00

00
01
02
03

00
00
00
00

TO
TO
TO
TO

00
00
00
00

00
00
00
00

FF
FF
FF
2E

TO
TO
TO
TO

12
12
12
12

The SCU selects the drive, moves access to the proper


cylinder, and selects the proper head. Any access motion
required is initiated after the seek address has been
transferred. Limitations on address byte values are:

Channel End is presented after the seek address has


been transferred. Device End is presented with Channel
End if no movement is required, or after access is
positioned if movement is required.
DATA ADDRESS-Specifies main storage location of the
seek address. Checked for both validity and correct parity
by the SCU. If an invalid seek address or parity error is
noted, the command is not executed; Unit Check,
Channel End, and Device End are presented in ending
status; and a subsequent sense command indicates
either command reject (invalid condition) or bus-out
parity error (parity error).
FLAGS-Can be used at discretion of the programmer.
COUNT-Must be six or greater. If less than six, command is not executed; Unit Check, Channel End, and
Device End are presented in ending status; and a
subsequent sense command indicates command reject.
INITIAL STATUS-Normally zero.

Bytes 0, 1, and 4 must be zero.

Bytes 2 and 3 together must not exceed 410


(dec) for 3670 or 814 (dec) for 3675.

Command execution does not require preceding CCW.

Byte 5 must not exceed 18 (dec).

File mask must be set to allow seeks.

(Figure 2-3)

SEEK HEAD
Command Code 1 B (hex)

Transfers the six-byte seek address shown below to the


SCU.

ADDRESS BYTES (HEX)


CYLINDER

BIN

HEAD

BYTE
0

BYTE
1

BYTE
2

BYTE
3

BYTE
4

BYTE
5

3670

00
00

00
00

00
01

00 TO FF
00 TO 9A

00
00

00 TO 12
00 TO 12

3675

00
00
00
00

00
00
00
00

00
01
02
03

00
00
00
00

00
00
00
00

00
00
00
00

TO
TO
TO
TO

FF
FF
FF
2E

The SCU selects the drive and the


Limitations on address byte values are:

TO
TO
TO
TO

12
12
12
12

DATA ADDRESS-Specifies main storage location of the


seek address. Checked for both validity and correct parity
by the SCU. If an invalid seek address or parity error is
noted, the command is not executed; Unit Check,
Channel End, and Device End are presented in ending
status; and a subsequent sense command indicates
either command reject (invalid condition) or bus-out
parity error (parity error).
FLAGS-Can be used at discretion of the programmer.
COUNT-Must be six or greater. If less than six, command is not executed; Unit Check, Channel End, and
Device End are presented in ending status; and a
subsequent sense command indicates command reject.

proper head.
INITIAL STATUS-Normally zero.

Bytes O. 1, and 4 must be zero.

Bytes 2 and 3 together must not exceed 410


(dec) for 3670 or 814 (dec) for 3675.

Command execution does not require preceding CCW.

Byte 5 must not exceed 18 (dec).

File mask must be set to allow seeks.

SPECIAL REOUIREMENTS-

SPECIAL REOUIREMENTS-

Channel End is presented after the seek address has


been transferred. Device End is presented with Channel
End if no movement is required, or after access is
positioned if movement is required.

2-8

SPACE COUNT
Command Code OF (hex)

(Figure 2-4)

Allows bypassing a defective count field on a track for


recovering data in key and/or data fields followtng the
defective field. Three bytes of information data are
transferred from the channel. These bytes are interpreted
by the SCU as the key length (first byte) and the data
length (last two bytes) of the record to be recovered.
DATA ADDRESS-Specifies main storage location of the
key and data lengths of record to be recovered.
FLAGS-Used at discretion of the programmer.

COUNT-Must be three or greater to transfer the requisite three bytes. If count is than three, the specified
number of bytes is transferred and the value of the
nontransferred bytes is assumed to be zero
CHAINING REQUIREMENTS

Cannot be chained from a Format Write or Erase.

Must not be followed by a Write, Erase, or Set File


Mask in the same chain.

304

~--

This command is used to space over a


bad count field. On the space count
command three bytes of data (key
length, 2 data length) are transferred
to the SCU

SPAC'COUNT
The data is used on the next command
to define the key and data length
because they could not be read fr'Jm
the count field. If the number of bytes
is less than 3. the succeeding counts
are set to zero.

Violation of these requirements will cause Channel End,


Device End, and Unit Check to be presented to the
channel.

037 (041)

LOCATE START
OF NEXT CNT FLO

This places the head at start of RO


count field.

CLOCK OVER G1,


HOME AODR
AND G2

037 (060)

SPACE OVER CNT


FLO. RECEIVE
3 BYTES OF DATA
FROM CHNL

037 (060)
SET END OF CNT
FLO ORIENTATION
FOR USE WITH
NEXT CMD

306
PRESENT CHNL
END AND DEVICE
END STATUS
TO CHNL

307 302

Figure 2-4. Space Count Flow Diagram

3672.21-0001-10175

2-9

RECALIBRATE
Command Code 13 (hex)

(Figure 2-5)

Causes addressed drive to seek to cylinder zero/head


zero. Channel End presented in ending status. Device
End presented when drive accesses to cylinder
zero/head zero.
DATA ADDRESS-Not checked for validity, but should
not exceed addressing capacity.

FLAGS-SLI flag must be set to avoid an Incorrect length


Indication
COUNT-Must not be zero. A zero count sets the
Program Check bit (bit 42) In the CSW
INITIAL STATUS-Normally zero. Not processed as an
immediate command.

3---

306

A'CAUBAAT>

This command causes the SCU to send


the rezero line to the drive. This causes
the access to move to cylinder zero.
The head address is reset to zero.

SPECIAL REQUIREMENTS-File mask must be set to


allow seek commands.

SEND CHNL END


STATUS TO CHNL

r--SCU- - -*'
I
I

DISCONNECTS
FROM CHANNEL

(IF

It
I

~~~~~E~)UX I
L _ _ _ ...;,..J

RST CTRL TAGS

MOVE ACCESS
IN REVERSE TO
CYL 000

032 (03E)

TO CONTROLLER

032 (03E)

START ACCESS
MOVEMENT IN
REVERSE
DIRECTION

DRV SENDS
AnENTION
TO SCU

SEND REZERO
TO CONTROLLER

.-- --,

SCU RECONNECTS
TO CHANNEL

(IF ~~~~~E~)UX

It

I
I

L ____ ..J
YES

306

306 If not Block Multiplex


302. 303 if Block Multiplex
If 2860 Attachment
Feature installed.
See paragraph 3.12.

306

SCU SENDS
DEVICE END
TO CHNL

SCU SENDS
CHNL END AND
DEVICE END
TO CHANNEL

307. 302

Figure 2-5. Recalibrate Flow Diagram

3672.21 -0001-10175

2-10

RESTORE

(Figure 2-6)

Command Code 17 (hex)

DATA ADDRESS-Not checked for validity, but must not


exceed addressing capacity
FLAGS-SLI flag must be set to avoid an Incorrect length
indicatIOn

Used primarily for compatibility with other direct-access


storage devices. It causes no action to be performed but
does result in loss of orientation. Channel End and
Device End immediately follow initial status

COUNT -Must not be zero A zero count sets the


Program Check bit (bit 42) In the CSW
INITIAL STATUS-Normally zero

This command does not perform any


operation.

RESTORE

304
RESET
TRACK
ORIENTATION

304 (05F)

PRESENT ZERO
STATUS TO CHNl

306

PRESENT CHNl END


AND DEVICE END
TO CHNl

307.302

Figure 2-6. Restore Flow Diagram

3672.21-0001-10175

2-11

SET FILE MASK


Command Code 1 F (hex)

(Figure 2-7)

File Mask command permits seek and write commands


(except Write Home Address and Write Record Zero).

Sets the write and seek masks which provide protection


for 3672-based subsystem data and defines Command
Retry-Program Control Interruption (PCI) interaction.
Defined values of the mask byte and their function are
listed below.

Channel End and Device End are presented to the


channel after the mask byte is transferred.

Bits 2 and 6 of the mask byte must be zero. If these bits


are not zero, the mask byte is considered to be invalid
and a Unit Check condition (CSW bit 38) is generated. A
subsequent sense command indicates Command Reject
Write or seek commands that violate the file mask are
not executed. In these cases, Unit Check is presented in
initial status, and either Command Reject (if a write
command) or File Protected (if a seek command) is
indicated by a subsequent sense command. Multitrack or
overflow operations that violate the file mask indicate
Unit Check and File Protected.
A system or selective reset resets the file mask to zero. A
Start 110 Instruction issued after a reset without a Set

DATA ADDRESS-Specifies main storage location of


mask byte.

-c

This command requires one byte of informa-

tion to be transferred to the SCU. This byte


defines

the

type

of

Write

and

Seek

operations that are allowed in a chain of


commands.

SET
FILE MASK

FLAGS-Used at discretion of the programmer.


COUNT-One.
INITIAL STATUS-Unit Check is presented if the file
mask is violated or more than one Set File Mask appears
in a chain.
CHAINING REQUIREMENTS-Only one Set File Mask is
permitted in a CCW chain. An attempt to issue more than
one Set File Mask in a chain causes Unit Check in initial
reset status, Command Reject to be noted in a subsequent sense command, and reset of the file mask to zero
at the end of the chain.

STORE FILE MASK


IN SCU BUFFER

306

036

Bit

5 6 7

a
a
a
[TI0
a a a a a a a
0
0

0 a 0 0 a
a 0 0 0 0 0
a a 0 a 1 a a
0 a a 1 a a 0
a a a 1 1 a 0
0 0 a 0 0 0 0
a 0 a 0 0 1 o
a 0 a 0 a 0
a 0 a 0 0 0 o

01
1
1 1

0
0

a
a
a
0
0

rn
1

Function:
Inhibit write HA and write RO.
Inhibit all write commands.
Inhibit all format write commands.
Permit all write commands.
Permit all seek commands.
Permit seek cylinder and seek head.
Permit seek head.
Inhibit all seek commands and head switching.
Inhibit diagnostic write command.
Permit diagnostic write commands.
Not PCI fetch mode.
PCI fetch mode. (The storage control presents UNIT CHECK if command retry is
used to recover from ECC uncorrectable data errors.)

IND THAT A FILE


MASK HAS BEEN
SET

306
SEND CHNL END
ANO DEVICE END
TO CHNL

307.302

Set to zero.

Figure 2-7. Set File Mask Flow Diagram

3672 .21 -0001-10175

2-12

SET SECTOR
C,lllll11,lI1d C"dt' 23 (hex)

(Figure 2-8)

Sector Number" 255


Command is treated as No-op. Channel End and
Device End are presented in ending status. Track
orientation is lost.

TI ,II1Sit'l S a sector Ilulllber (0-127 dec) from main storage


Itl tilt' SCU. This command is used on block multiplex
cll;lI1lwls Itl elllllinate the need to maintain channel and
SCU ClllllWctltl1l during drive rotational delays.
TIle sector number is checked for validity by the SCU.
One of three validity conditions results: valid sector
number (0-127), sector number > 127 and < 255, or
sector number" 255.

SET SECTOR

DATA ADDRESS-Specifies main storage Iqcation of


desired sector number.

038 (03C)

FLAGS-Used at discretion of the programmer.

RECEIVE SCTR
NO. BYTE
FROM CHNL

One of 128 positions


per track.

Valid Sector Number (0-127)


SCU presents Channel End and disconnects.
Device End is presented when sector number is
reached and channel reconnects to continue chain.
If reconnection does not occur, the SCU attempts
to reconnect on subsequent revolutions. (All valid
sector numbers are adjusted by the SCU to compensate for channel reselection delay.)

This command requires one byte of


information to be transferred to the
SCU. The byte transferred gives the
sector number requested by the
channel. The SCU sets up the 3670
Drive to signal when the requested
sector is approaching the read/write
heads. The SCU raises REQUEST-IN to
the channel. If the channel does not
respond within 130 f,lSec, the SCU
drops REQUEST-IN and waits until the
drive signals the sector arrival on the
next revolution.

COUNT-One.

SPECIAL REQUIREMENTS-

Set Sector does not guarantee record orientation.


Search commands must still be used for this function.

127 ' Sector Number < 255

Channel, Device End, and Unit Check are


presented in ending status. Command Reject indicated in a subsequent sense command.

Indiscriminate use of Set Sector with multitrack


search may result in missing the desired record. A Set
Sector 0, Read Home Address, Search Multitrack
sequence will prevent this condition from occurring.

CHNL RESPONSE

303
PROCESS ANY
DRV INTS
038 (031)
SCU compensates the sector
value to ensure that
reconnection occurs at
proper value.

SEND SCTA NO.


TO CONTROLLER

-----..

306

SEND DEVICE END


TO CHNL
306

r--

--.,

: SCU DISCONNECTS II
I FROM CHANNEL I

L__

If 2860 Attachment Feature


stalled. See paragraph 3.12.

__..J

in-

Figure 2-8. Set Sector Flow Diagram

3672.21-0001-10/75

2-13

DIAGNOSTIC LOAD
Command Code 53 (hex)

(Figure 2-9)

Transfers a G68-byte block of data addressed by a control


byte from the flexible disc file to writable control storage
an the SCU. The data block transferred is a functional
microprogram diagnostic test. During the transfer, the
SCU is disconnected from the channel.
The control byte, specifying the diagnostic 10 control
number. IS transferred from main storage to the SCU.
Bits 0-5 of the control byte specify the flexible disc track
address (0-49 dec) and bits 6 and 7 specify the sector
pair (0-3).
The SCU checks the control byte for validity. A valid
control byte presents Channel End in ending status. An
invalid control byte sets Unit Check in the status byte and
Command Reject in a subsequent sense byte.

Completion of the data transfer causes the SCU to


request service and present Device End when polled.
DATA ADDRESS-Specifies main storage location of
control byte.
DIAGNOSTIC
LOAD

FLAGS""':Used at discretion of the programmer.

HEX 53

COUNT-One.
INITIAL STATUS-Normally zero.
NO

SPECIAL REQUIREMENTS

Command execution allows any drive address to be


used with the SCU address.
XFER
1 BYTE
TRK/SCTR
ARGUMENT
FROM CHNL

Read Diagnostic Status 1 command transfers the


diagnostic test from writable control storage in the
SCU to main storage.

0123467

NO

SET CE, DE,


AND UC STATUS.
SET
INTERVENTION
REQUIRED SENSE.
PRESENT
CE, DE, AND
UCSTATUS
TO CHNL

PRESENT CE
ENDING STATUS
TO CHNL

RAISE REQUEST-IN
TO PRESENT DE
STATUS TO CHNL

Figure 2-9. Diagnostic Load Flow Diagram

2-14

DIAGNOSTIC WRITE
Command Code 73 (hex)

(Figure 2-10)

Transfers a 668-byte diagnostic test from main storage to


writable control storage in the SCU. Upon completion of
the data transfer, execution of the test begins. At
completion of the test, a 16-byte error code message is
stored in buffer storage of the SCU. The message is
transferred from buffer storage to main storage by a
subsequent Read Diagnostic Status 1 command.
Channel End is presented after transfer of diagnostic test
to the SCU. Device End is presented after the test is
complete.

DATA ADDRESS-Specifies main storage location of the


diagnostic test.

---1. __

FLAGS-Used at discretion of the programmer

H_E_X_7_3_ _

COUNT-Should be 668. If greater than 668, only 668


bytes are transferred. If less than 668, only the specified
number of bytes are transferred; command is terminated;
and Channel End, Device End, and Unit Check are
presented in ending status.

INITIAL STATUS-Normally zero.

PRESENT
CE STATUS
TO CHNL

SET CE, DE, AND


UC STATUS.
SET INTERVENTION REQD
IN SENSE BYTES

EXECUTE DIAG
AND ASSEMBLE
RSLTS

PRESENT CE, DE,


AND UC STATUS
TO CHNL

RAISE REQUESTIN
TO PRESENT DE
STATUS TO CHNL

Figure 2-10. Diagnostic Write Flow Diagram

2-15

2.5 SENSE COMMANDS

Of the 24 bytes of sense information. the bit-by-bit


meaning of messages reported in bytes 0 through 6 is
preassigned. See Figure 2-11. The meaning of messages
reported in bytes 8 through 23 depends on the format
used for these bytes. This format is decoded in byte 7
(Bits 0 through 3).

2.5.1 Sense Byte Information


Conditions that occur during the execution of an instruction command sequence are reported to the using system
program by the channel status word (CSW) and sense
information. The sense information defines general and
specific conditions in both the SCU. controller, and the
drives that are not defined in the CSW. The sense
information is contained in sense bytes which are
transferred to the using system by the Sense I/O
command. Upon receiving a Sense I/O command, the
SCU transfers the 24 bytes of sense information to the
using system. This command is normally executed
following a status byte containing Unit Check (CSW bit
38)

The tables that follow this paragraph describe how sense


bytes are assembled, the bit-by-bit meaning of each
sense byte, and the formats and messages associated
with sense bytes 8 through 23.

Bit 0 (Write Inhibit) turns on bit 6 of sense byte

Bit 1 (Operation Incomplete) turns on bit 7 of


sense byte 1 and defines a Restart command in
sense byte 3.

Bit 2 (Correctable) turns on bit 1 of sense byte 2.

Bits 5, 6, and 7 specify other bits which may be


turned on in addition to those described above.
The byte in which a bit is turned on is specified
by bits 3 and 4.

The Format/Message byte is formatted in the same


manner as sense byte 7.

2.5.2 Sense Control Block


The sense control block (Table 2-2.) is a set of two bytes
which determine which bits are to be turned on in the
sense assembly. The Sense Code byte sets bits in sense
bytes O. 1, and 2 as follows:

2.5_3 Sense Byte Bit Definitions

1.

Upon entry. the sense bytes defined by the sense control


block are stored by the subroutine at the starting location
in G23.

Definitions of bits in all 24 sense bytes are listed in


Tables 2-3. 2-4. and 2-5. Table 2-3 defines bits for sense
bytes 0 through 6 Table 2-4 lists the definitions for
format/message byte 7. Table 2-5 defines bits for bytes 8
through 23.

2.5.4 Assembling Sense Bytes 8 through 23


Sense bytes 8 through 23 are the second bytes to be
assembled. These bytes are assembled according to the
format specified in bits 0 through 3 of sense byte 7. The
bit-by-bit meaning of sense bytes 8 through 23 for
formats 0 through 6 are given in Table 2-5.

2.5.5 Command Descriptions


Each sense command is described on the following
pages by means of a flow diagram and a narrative
description.

TABLE 2-2. SENSE CONTROL BLOCK

BIT

SENSE BYTES

1 1 1 2

3 1 4 I 5

I6

7 I 8

I9

1 10 1 11 112 1 13 1 14 1 15 1 16 117 I 18

j.--

J 19 120 1 21 J 22 J 23

DESCRIBE THE ERROR


CONDITIONS IN DETAILED TERMS
CONTAINS FORMAT AND MESSAGE
DECODES OF BYTE 8-23

--

SENSE CODE

FORMATIMESSAGE

Write Inhibit

Bits 0 through 3
are decoded into
formats 0 through 6.
Formats 0 through 6
are defined as follows:
0 Message Only
1 Device Errors
2 SCU Errors
3 Selective Reset
4 ECC Uncorrectable
5 ECC Correctable
6 Usage/Error Count

Operation Incomplete

Correctable

Bit
Arrangement:
01
10
11
00

DESCRIBE THE INITIAL


CONDITIONS

DESCRIBE THE ERROR


CONDITIONS IN GENERAL TERMS

f--

Figure 2-11. Sense Byte Definitions

6
7

o and bit 0
of byte 1

Byte Number:
0
1

Bit Position
in Byte

Bits 4 through 7
are decoded into
Message 0 through C.
The meaning of the
message depends upon
the format decoded.

2-16

TABLE 2-5. SENSE BYTES 8-23 BIT DEFINITIONS (1 OF 2)

FORMAT O-PROGRAMMING OR SYSTEM CHECK

SENSE BYTE 13

SENSE BYTES 823


NOT USED SET TO ZERO

Contents of DO Register

SENSE BYTE 19-CONTROLLER CHECK 19


Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

SENSE BYTE 14
FORMAT 1-DISC DRIVE EQUIPMENT CHECK

Contents of 01 Register

SENSE BYTE 8-MODULE STATUS

Bit 1
.,,0
Bit
2
Bit 3
Bit 4
Bit 5
Bit
Bit 7

01

:::~

2
3
4
5
6
7

}
01

Index error
Offset active
Seek Incomplete
Seek complete
Online
Attention
Busy
Record ready

Bit 1

Bit
Bit
Bit
Bit

4
5
6
7

0'

Not used
DiagnostiC 4
Diagnostic 2
Diagnostic 1
Not used
Mode 4
Mode 2
Mode 1

ORDER 2C
120
DO

Monitor
Monitor
Monitor
Monitor
Monitor
Monitor
Monitor
Monitor

state
state
state
state
state
state
state
state

8
7
6
5
4
3
2
1

CE program ,t()[l
Not used
Interface check
Monitor check
Not used
Drive command reJect

SENSE BYTE 12--SAFETY


Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

0
1
2
3
4
5
6
7

0'

ORDER = 04
DO = 104

Data Transfer Counter Low Byte

SENSE BYTES 20-DRIVE INOP 2

SENSE BYTE 16-CONTROLLER CHECK 16


Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

0
1
2
3
4
5
6
7

Not used
Write parity error
Read parity error
Bit ring error
Write compensation error
Data transfer control error
Missing PLO pulses
VFO phase error

ORDER 2C
DO = '10

Select lock
Not used
Abnormal stop
Not used
Servo disable
Seek not enabled
Not servo data
Even cylinder

ORDER 04
DO' '02

ECC no Input data


ECC PO or write error
ECC Pl or P3 error
ECC P2 error
ORDER 04
Sync out check
DO = 101
PLO reorient counter check
Gap counter check
Gap control check

1
2
3
4
5
6
7

SENSE BYTE 13

Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

0
1
2
3
4
5
6
7

Write Overrun
ACWrite
Write Fail
Subtractor Fail
Multi Write
Multi Head
35V Reg. Fail
Multi Write Data (3670)
Pad Unsafe (3675)

SENSE BYTE 14
ORDER = 2E
DO = 112

Contents of 01 Register
SENSE BYTE 15
Contents of CO Register
SENSE BYTES 16-21
SET TO ZERO

DRIVE INOP 3

Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

Write Offset
Write & No Fine Track
Read & Write
Short Write
AC Write S.S.
Add Mark S.S.
Sink Fail
Not used

SENSE BYTES 22-23-ERROR SYMPTOM CODE

1
2
3
4
5
6
7

FORMAT 3-SELECTIVE RESET

SENSE BYTE 1B-CONTROLLER CHECK 18

ORDER 2C
DO = 108

0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100-1111

Controller check
MICROSelect active chec~
PROGRAM
GENERATED
01 check
Reserved
following errors:'
Not used
Tag valid missing (read. write)
Normal end check end missing (control)
Normal end miSSing (control)
No index after 40 ms or solid index
Unexpected status with check end
Controller selection address check
Preselection check
Zero pattern alignment check
Repetitive command overrun
Drive interrupt during busy
Drive status not as expected after
Seek or Set Sector command
Not used

SENSE BYTE 8

ORDER = 2E
DO = 113
Bit 1
Bit 2
'"0

SENSE BYTE 22 AND 23-ERROR SYMPTOM CODE

Bit
Bit 1
Bit 2
Bit 3
Bits 4-7 define the

MCM = 138

Contents of DO Register

SENSE BYTE 21

Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

MCM" 28

SENSE BYTE 11

SENSE BYTE 17--CONTROLLER CHECK 17

SENSE BYTE l1-CHECK STATUS


Bit 0
Bits 1 -3
Bit 4
01
Bit 5
\
Bit 6
Bit 7

Data Transfer Counter High Byte

SENSE BYTE 12
SET TO ZERO

Contents of CO Register
ORDER 24
DO = 100

SENSE BYTE 10-MONITOR STATE

.,0
Bit 2
.0< 3

Drive selection error


CTLI tag bus check
Device check
CTL-I bus out check
Write sense check
Read or write valid check
Device bus out check
Controller bus in assembler check

STATUS BYTE 15

SENSE BYTE 9-MONITOR MODE

Bit
Bit
Bit
BII
Bit
Bit

0
1
2
3
4
5
6
7

SENSE BYTE 10

FORMAT 2-SCU ERRORS

Bit
Bit
Bit
Bit
Bit

3
4
5
6
7

Bit

MCK

Bit 1
Bit 2
'0<
Bit
Bit
Bit
Bit
Bit

3
4
5
6
7

ECR

MCM = 108

.,,' ~
Bit
Bit
Bit
Bit
Bit
Bit

2
3
4
5
6
7

MCK

Bit

"" ~
2
3
4
5
6
7

ECR

01 Buffer Check
Select Active Check
Not Used
CE Alert
Channel Buffer Read Check
Channel AIC Interface Check
Channel BID Interface Check
Channel Transfer Error

1
2
3
4

MCM = 103

Failing
Failing
Failing
Failing
Failing
Failing
Failing
Failing

Address
Address
Address
Address
Address
Address
Address
Address

Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

5
6
7
8
9
10
11
12

MCM = 104

SENSE BYTE 10

SENSE BYTE 9-SERDES CHECK

Bit
Bit
Bit
Bit
Bit
Bit

SENSE BYTE 9

SENSE BYTE 8-CONTROL CHECK


Select Alert 1
Select Active
Sync In
Unexpected End Check
Normal End
Check End
Tag Valid
Not Used

Selective Reset
Single Buffer Error
Single WCS Error
Failing Address Bit
Failing Address Bit
Failing Address Bit
Failing AddreSS Bit
Failing Address Bit

Bit 1
Bit 2
,"0
MCM = 118

Bit
Bit
Bit
Bit
Bit

3
4
5
6
7

MCK

Not Used
Multi-WCS Error
A-Bus
B-Bus
T-Bus
Branch
IMPL
ALU

MCM = 110

2-19

TABLE 2-5. SENSE BYTES 8-23 BIT DEFINITIONS (2 OF 2)

SENSE BYTE 11

~O:

Bit
Bit
Bit
Bit
Bit
Bit
Bit

1
2
3
4
5
6
7

MCK

BR Multiplex Error
PS Error
OP Code Translation Error
Subroutine Error
WCS Error
Multiplexer Buffer Error
Buffer ECC Error
8uffer Write T-8us Error

FORMAT 4-DATA CHECKS NOT PROVIDING


DISPLACEMENT INFORMATION

FORMAT5-DATA CHECKS PROVIDING DISPLACEMENT


INFORMATION

FORMAT 6-USAGE/ERROR COUNTS

SENSE BYTE 8-CYLINDER (1 )

SENSE BYTE 8-CYLINDER (1 )

SENSE BYTES 8- I I-BYTES READ

High-order cylinder byte of last seek address

High-order cylinder byte of last seek address

These four bytes provide an accumulated count olthe number of bytes


processed by the controller in read operations.

SENSE 8YTE 9-CYLINDER (2)

SENSE BYTE 9-CYLINDER (2)

SENSE BYTE la-HEAD (1)

Bit I
Bit 2

~"45

Bit
Bit
8,t
Bit

. Me<

6
7

Failing
Failing
Failing
Failing
Faoling
Failing
Falling
Failing

PI
P2
P3
P4
P5
P6
P7
P8

Low-order head byte of last seek address


SENSE BYTE 12-RECORD
Record number of record in error

Sector number of record In error

Contents of CHF Register

SENSE BYTE 14-0FFSET


Amount of offset used to recover from error

SENSE BYTE 14
Conents of RWC Register

MCK

High-order head byte of last seek address

SENSE BYTES 14 AND 15-UNCORRECTABLE DATA CHECKS

SENSE BYTE 11 -HEAD (2)

These two bytes identify the number of uncorrectable data checks


which have. been recorded.

DRIVE
DATA
WORD 4

Low-order head byte of last seek address

SENSE BYTE 13

1
2
3
4
5
6
7

These two bytes provide an accumulated count of the number of ECC


correctable data checks which have been recorded.

SENSE BYTE II-HEAD (2)

Bit
Bit
Bit
Bit
Bit
Bit
Bit
Bit

SENSE BYTE 13-SECTOR

Bit
Bit
Bit
Bit
Bot
Bit
Bit

SENSE BYTE 10-HEAD (1)

High-order head byte of last seek address

MCM,102

, .I

Low-order cylinder byte of last seek address

BUFFER
LOC L
13B

MCM = 101

,... ~

DRIVE
DATA
WORD 3

SENSE BYTES 12 AND 13-CORRECTABLE DATA CHECKS

Loworder cylinder byte of last seek address

SENSE BYTE 12

1-

}RO;C

Number of retries required to recover from error


SENSE BYTE 16-S0URCE DRIVE IDENTIFICATION
GA Error
GB Error
GC Error
GD Error
FD Read Error
FD Seek Error
FD Not Ready
Write Bus Error

Bits 0 and 1

Bits 2-7

Identifies the controller that was used to record the


data in which the error occurred

Identifies the disc drove that was used to record the


data in which the error occurred.

MCM = 105
SENSE BYTES 16-21
SET TO ZERO

SENSE BYTES 17-21


NOT USED-SET TO 0

SENSE BYTES 22 AND 23


ERROR SYMPTOM CODE

SENSE BYTES 22 AND 23


ERROR SYMPTOM CODE

}~ ~;E3R

:\

BUFFER
LOC
137

BUFFER
LOC
12E

SENSE BYTE 18

Bits 1-7

If zero. bytes 20-23 contain channel A and B information; if one. channel C and D information.
Not Used

f-

DRIVE
DATA
WORD 5

Amount of offset used to recover from error


SENSE BYTES I 5-17-RESTART DISPLACEMENT

Specifies the number of bytes processed by the storage control to end


of data field in error

BUFFER
12D

SENSE BYTE 13-SECTOR

Bit 0

BUFFER
LOC 13
BYTE 2

LOC

These two bytes provide a count of the number of access motions


initiated by the channel.

SENSE BYTE 14-ACCESS OFFSET

BUFFER
LOC 13
BYTE 3
}

Record number of record in error

Sector number of record in error

BYTES
OAND

SENSE 8YTE 15-RETRIES

SENSE BYTE 15

SENSE BYTES 16 AND 17-SEEKS


SENSE BYTE 12-RECORD

SENSE BYTE I 9-SEEK ERRORS


Identifies the total number of seek error which were successfully retried by the storage control.

SENSE BYTES 18 AND 19-ERROR DISPLACEMENT


Displacement offirst byte in error relative to end of the data field where
error occurred
SENSE BYTES 20-22-ERROR PATIERN

SENSE BYTE 20 AND 21-COMMAND AND DATA OVERRUN.


CHANNEL A OR C
Provides a count of the number of command overruns (byte 20) or data
overruns (byte 21) which were retried by the storage control for
channel A or C.

1
BUFFER

LOC
Contain high. middle. and low error pattern bytes used for error correction function.

SENSE BYTE 22 AND 23-COMMAND AND DATA OVERRUN.


CHANNEL B OR D

127 AND
128

SENSE BYTE 23
Bits 0-6
Bit 7

Not used-set to 0
Channel truncation

Provides a count of the number of command overruns (byte 22) or data


overruns (byte 23) which were retried by the storage control for
channel B or D.

2-20

SENSE 110
Command Code 04 (hex)

(Figure 2-12)

Transfers 24 bytes of sense Information from the SCU to


the channel This Information describes unit check
status. current status of the device that performed the
operation, and system error recovery information.
Channel End and Device End are presented after sense
bytes are transferred.
DATA ADDRESS-Specifies main storage location where
sense bytes are to be transferred.

FLAGS- .Used at discretion of the programmer


304 (05F)

COUNT---24

INITIAL STATUS-Normally zero.

SPECIAL REQUIREMENTS-Unit Check should always be


followed by a Sense command whether or not sense
information is used; otherwise, expected future interrupts may not occur and some I/O access paths may
be unavailable.

SENSE I/O

YES

014 (OIA)

XFER SENSE BYTES,


STORED IN BUFFER
LOCATION 120125,
TO CHNL UNTIL
ALL ARE TAKEN
OR TRUNCATION
OCCURS. ZERO
OUT SENSE AREA
DURING XFER,

XFER SENSE BYTES,


STORED IN BUFFER
LOCATION 131136,
TO CHNL UNTIL ALL
ARE TAKEN OR
TRUNCATION
OCCURS, ZERO
OUT SENSE AREA
DURING XFER.
RST LOG POINTER
IN UNIT CHECK
OWEDWRD,

306

PRESENT
CE, DE TO
CHANNEL

307,302

Figure 2-12. Sense I/O Flow Diagram

3672.21 -0001-10175

2-21

TeST I/O
Command Code 00 (hex)

(Figure 2 -13)

Transfers one status byte from the SCU to the channel.


This byte is normally zero except when stacked or
pending status is presented for the SCU or disc drive.
Then, the status byte records status of the SCU and disc
drive on the channel.

DATA ADDRESS--Specifies main storage location where


status byte is to be transferred.

FLAGS-Not used.

COUNT -1.

3--

TEST I/O

NOTE
This command is a system instruction,
not a CCW. The subsystem treats the
command as an immediate comm'and.
One status byte is returned, to the
system.

306

PRESENT STATUS
TOCHNL

306

PRESENT ZERO
STATUS TO CHNL

307

Figure 2-13. Test I/O Flow Diagram

3672.21-0001-10/75

2-22

READ AND RESET BUFFERED LOG


Command Code A4 (hex)

(Figure 2-14)

Transfers 24 bytes of usage or error information from the


SCU to the channel. This information. generated and
available when their respective counters overflow. pertains to the SCU addressed by the Start //0 instruction
and the drive identified in sense byte 4. The counters are
reset after the data transfer.
Channel End and Device End are presented after the data
transfer.

DATA ADDRESS-Specifies main storage location of first


error byte or usage information.

FLAGS-Used at discretion of the programmer.

COUNT-24.

INITIAL STATUS-Normally zero.

3-

READ AND RESET


BUFFER LOG

014 (07C)
SET UP SENSE INFO IN
BUFFER LOCATION
/0120. SET SCB AS
FOLLOWS:
A. FMT = 6
B. MSG = 0
C. ENVIRONMENTAL
DATA PRESENT
014 (alA)
XFER SENSE BYTES.
STORED IN BUFFER
LOCATION 120125.
TO CHNL UNTIL ALL
ARE TAKEN OR
TRUNCATION OCCURS.
ZERO OUT SENSE
AREA DURING XFER.

306

PRESENT CEo DE

GO TO
RESELECTION
ROUTINE

307

Figure 2-14. Read and Reset Buffer Log Flow Diagram

3672.21-0001 -10175

2-23

READ DIAGNOSTIC STATUS 1


Command Code 44 (hex)

(Figure 2-15)

This command may perform either of two functions,


depending on whether it follows a Diagnostic Write or a
Diagnostic Read command. If it follows a Diagnostic
Write, the command transfers a 16-byte error code
message from buffer storage in the SCU to main storage.
For this type of transfer. the CCW Count field should
specify 16. Channel End and Device End are presented
after the transfer.
If it follows a Diagnostic Load. the command transfers a
668-byte diagnostic test from buffer storage in the SCU
to main storage. For this type of transfer, the CCW Count
field should specify 668. Channel End and Device End
are presented. after the transfer.

DATA ADDRESS-Specifies main storage location where


data accumulated during prior Diagnostic Load or
Diagnostic Write is to be stored.
FLAGS-Used at discretion of the programer.
READ
DIAGNOSTIC
STATUS 1

COUNT-16 or 668, depending on type of transfer.


INITIAL STATUS-Normally zero.
SPECIAL
Diagnostic
otherwise,
storage in
message.

REQUIREMENTS-Diagnostic Load or
Write. must precede Read Diagnostic Status 1;
16 bytes of data are transferred from buffer
the SCU which normally contain the error

NO

SET CE, DE, AND


UC STATUS.
SET INTERVENTION REQ'D
IN SENSE.

NO

XFER 16 BYTES
OF ERROR CODE
TO CHNL

XFER
668 BYTES
TO CHNL

28 bytes of control information and


640 bytes of microprogram statements.

SET CE AND
DE STATUS

Figure 2-15. Read Diagnostic Status 1 Flow Diagram

2-24

DEVICE RESERVE
Command Code B4 (hex)

(Figure 2-16)

Reserves the addressed drive to the channel issuing the


command Reservation is maintained until either a
DevIce Release or a system reset is performed by the
channel. In addition, 24 bytes of sense information are
transfered to main storage across the channel.
Normal busy conditions cause a command reject and set
the CSW Busy bit. Abnormal file status conditions (file
unsafe, off-line. and so forth) do not halt command
execution
Channel End and Device End are presented after sense
byte transfer.

DATA ADDRESS-Specifies main storage location where


sense bytes are to be transferred

FLAGS-Used at discretion of the programmer


COUNT

24

INITIAL STATUS-Normally zero.

HEX B4
RESERVE

FROM INITIAL
SELECTION

304

DEVICE RESERVE
84, DEVICE
RELEASE 94

HEX 94
RELEASE

SPECIAL REQUIREMENTS-Device Reserve may not be


executed if a Set File Mask precedes command in the
same chain.

YES

Violation of this requirement will cause command to be


rejected, set CSW Unit Check bit. and Command Reject
to be noted in a subsequent sense command.
SEND ZERO
STATUS
TO CHNL

304
NO (RELEASE CMD)

DEVICE RELEASE
Command Code 94 (hex)

(Figure 2-16)

Terminates reservation of the drive address reserved by


Device Reserve. In addition, 24 bytes of sense information are transferred to main storage across the channel.

DATA ADDRESS-Specifies main storage location where


sense bytes are to be transferred.
FLAGS-Used at discretion of the programer.

SET CMD REJECT


AND INVALID
SEa IN SENSE

078

011

RELEASE
DEVICE

RESERVE
DEVICE

COUNT-24.
INITIAL STATUS-Normally zero.

Normal busy conditions cause a command reject and set


the CSW Busy bit. Abnormal file status conditions (file
unsafe, off-line, and so forth) do not halt command
execution.

SPECIAL REQUIREMENTS-Device Release may not be


executed if a Set File Mask precedes command in the
same chain.

Channel End and Device End are presented after sense


byte transfer

Violation of this requirement will cause command to be


rejected set CSW Unit Check bit. and Command Sense
to be noted In a subsequent sense command

SET UNIT
CHK IN
CHNL STATUS

L--------I~ 014 (01A)

SEND 24
SENSE BYTES

306

307,302

Figure 2-16. Device Reserve and Device Release Flow Diagrams

3672 .21 -0001-1 0/75

2-25

2.5.6 Transfer In-Line Diagnostics to SCU


2.5.6.1 INTRODUCTION
To provide maximum facility availability, the SCU can
execute diagnostic tests on a drive concurrent with
normal system operations on the remaining drives. The
diagnostic tests are transferred from the system
diagnostic library in main storage by a Transfer In-Line
Diagnostics to SCU operation. This mode of operation
allows the FSR to diagnose and repair most drive failures
while the facility continues to operate other attached
drives.
After the diagnostic is run, the error code is returned to
the system. To permit temporary residence for a specific
diagnostic test, the SCU provides a transient block of 128
words of control storage. The transient area is loaded
under control of the On Line Test Executive Program
(OLTEP). The Diagnostic Write command loads a selected
test into control storage and instructs the SCU to execute
the test. This loading and execution can also be initiated
from the FE panel. After the test, error message informa-

tion or test results are transferred from the SCU to main


storage by a Read Diagnostic Status 1 command. If the
FE panel is used, the test results are displayed by the FE
panel indicators.

Transfers status of subsystem to the system.

Runs in-line diagnostic.

Transfers diagnostic data to the subsystem.

Stores 16-byte error code in the Control Storage


buffer area if errors are found.

Transfers error code from subsystem to the system

Sends Device End status to the channel.

The following text material is keyed to the block diagram


of Figure 2-17.

2.5.6.2 SYSTEM

Executes a Start I/O instruction which addresses a


Diagnostic Write command.

After receiving Device End status, executes a Start


110 instruction which addresses a Read Diagnostic
Status 1 command.

2.5.6.4 CHANNEL INTERFACE

Decodes Read Diagnostic Status 1 command.

Controls timing of data to and from the channel.

Transfers error code from the control storage to the


channel on Read Diagnostic Status 1 command.

Checks parity of data transferred to the SCU.

Sends Channel End and Device End status to the


channel.

2.5.6.5 STORAGE CONTROL UNIT (SCU)

Transfers address and command information to the


selected subsystem.

STATUS AND
ERROR MESSAGE
..... DATA

....

Stores main microprogram to control Diagnostic Write


and Read Diagnostic Status 1 commands.

Stores diagnostic data from system in a buffer area.

Sends Channel End to the channel.

Stores error code for any errors found.

CHANNEL
INTERFACE

CHANNEL

Sets up and controls the transfer of data from


Channel Interface to the Control Storage diagnostic
buffer area.

......

........

2.5.6.6 CONTROL STORAGE

Decodes the Diagnostic Write command.

2.5.6.3 CHANNEL

ADDRESS COMMAND
AND DIAGNOSTIC
PROGRAM DATA
SYSTEM

.......

.....
....

SCU

I.....

DIAGNOSTIC
PROGRAM
DATA IN

~. DIAGNOSTIC
PROGRAMC ONTROl
OUT

1,
ADDRESS
CONTROl~

CONTROL STORAGE

Figure 2-17. Transfer In-Line Diagnostics to SCU Block Diagram

2-26

2.6 READ COMMANDS


2.6.1 Operation Block Diagram
Read commands transfer data from the selected disc
pack to the using system. These commands can operate
in multitrack mode, indicated by setting bit 0 of the
command code to one. Multi-track operations allow the
SCU to automatically switch heads at index time without
a Seek Head command.

Overflow records allow data records of more than one


track in length. Overflow operations are indicated by a bit
in the flag byte.
Data checks may occur during execution of a read
command. If such checks are detected, the SCU uses
ECC to correct the error if correctable, or Command Retry
if not correctable.

2.6.1.1 SYSTEM
The system executes the Start 110 instruction which
addresses a channel command word (CCW) containing a
read command.

USING SYSTEM

IIF

I
I

I
I

READ
DATA
CONTROL

CHNL

CHANNEL

SYSTEM

I
I
L __

Transfers the address of the selected data to the SCU.

Transfers data from the SCU to the using system.

Controls the timing of the transfer of data and control


information between the channel and the SCU.

Checks parity of data transferred to the SCU.

Decodes the read command.

Selects the addressed drive head.

2.6.1.6 READ CIRCUIT

Transfers ending status to the channel.

Sets up the drive to read data from the disc.

Indicates retry status to the channel if error is a type


that can be retried, and sets up to relocate record.

Changes currents from the read head into data pulses


to SERDES.

Recognizes gap configurations.

Provides the SCU with status of the drive.

2.6.1.5 CONTROLLER INTERFACE

2.6.1.7 READ/WRITE HEAD

Transfers read control information to the selected


drive.

Changes magnetic flux from pack to read current


signals.

Transfers status
Microprocessor.

Selected head provides data pulses.

Controls the serial-by-byte transfer of read data from


the serializer/deserializer (SERDES) to the controller
interface.

the

selected

drive

to

the

PLO head provides PLO pulses.

Servo head and circuits hold access at selected track.

2.6.1.S PACK

Receives the serial data from the drive.

Recognizes gaps and synchronizes to the data.

Changes serial-by-bit drive data to serial-by-byte data.

The pack contains bits of information on the magnetic


disc coating.

Checks the status of the drive and transfers the status


to the channel.

Sets up the read tags to the controller.

Sends data to the ECC and the Microprocessor.

2.6.2 Command Descriptions

Controls the byte-by-byte transfer of read data from


the controller.

Uses PLO pulses to maintain synchronization when


not reading.

Each read command is described in the following pages


by means of a narrative description and a flow diagram.

I
I
I

SCU

CHANNEL
INTERFACE

_ I TAG IN
-I
MICROPROCESSOR

READ
DATA
AND
STATUS

IIF

-CONTROLLER

I
I
1

: I INDEX

------I

I
SERVO
TRACK
.J>ATA

I READ CONTROL

-I

__ I SER RD DATA I
: 1PLO
I

,'"

I
I

I BUS OUT 1
,I BUS IN
-I

I-----~ISCDRI~------I

I
I

1TAG OUT I..

..

..

CTL

I
I

READ
DATA
CONTROL

1
READ
DATA
AND
STATUS

of

I
I

READ
DATA
CONTROL 1

_I
READ
DATA
AND
STATUS

Executes the CCW to transfer the read command to


the SCU.

Monitors incoming tags for error indicators.

2.6.1.4 MICROPROCESSOR

The following text material is keyed to the read operation


block diagram shown in Figure 2-18.

2.6.1.3 CHANNEL INTERFACE

The Read Count, Key, and Data; Read Key and Data; and
Read Data commands can operate on overflow records.

I
I

2.6.1.2 CHANNEL

DEVICE STATUS

READ
CIRCUIT

READ
DATA

--

SERVO
TRACK
_DATA

READ/WRITE
HEAD

I
I

PACK

READ
DATA

--

L ________________

Figure 2-18. Read Operation Block Diagram

2-27

READ DATA

(Figure 2-19)

Command Code: 06 (hex) single track


86 (hex) multi-track

Transfers the data area of a record from the disc to main


storage. The data read can be one of the following:

Data area of record read by Search 10 or Search


Key from which Read Data is chained.

Data area of record read by Read Count from


which Read Data is chained.

Data area of record following next count area on


the track (excluding RO).

Correction codes following each data area are used to


check for data validity. A parity bit is added to each byte
prior to its transfer to the channel.

Channel End and Device End are presented to the


channel upon completion of the correction code check of
the data area.
A data overrun or check condition, if detected, initiates
an SCU recovery attempt by Command Retry. If Command Retry is not used or is unsuccessful when an
overrun or check condition occurs, Channel End, Device
End, and Unit Check are presented to the channel.
(Command Retry is not used if a correctable data error;
i.e., error burst of 11 bytes or less, is detected in the data
area.)
DATA ADDRESS-Specifies main storage location where
first data byte is to be transferred.

READ KEY AND DATA

Transfers the key and data areas of a record from the


disc to main storage, The key and data read can be one of
the following:

Key and data area of record read by Search 10


from which Read Key and Data is chained.

Key and. data areas of record ready by Read


Count from which Read Key and Data is chained,

FLAGS-Used at discretion of the programmer.


COUNT-Specifies number of bytes to be read.
INITIAL STATUS-Normally zero.

(Figure 2-19)

Command Code: OE (hex) single track


8E (hex) multi-track

Key and data areas of record following next


count area on the track (excluding RO).

Correction codes following each key and data area are


used to check for data validity.. A parity bit is added to
each byte prior to its transfer to the channel.

Gl. G2, AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

Channel End ann Device End are presented to the


channel upon completion of the correction code check of
the areas,
A data overrun or check condition, if detected, initiates
an SCU recovery attempt by Command Retry. If Command Retry is not used or is unsuccessful when an
overrun or check condition occurs, Channel End, Device
End, and Unit Check are presented to the channel.
(Command Retry is not used if a correctable data error;
i.e., error burst of 11 bytes or less, is detected in the data
area.)
A key length of zero causes the command to operate as a
Read Data command.
DATA ADDRESS-Specifies main storage location where
first byte of key data is to be transferred.
FLAGS-Used at discretion of the programmer,
COUNT-Specifies the number of key and data area
bytes to be read.
INITIAL STATUS-Normally zero.

Gl, G2, AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

2-28

READ COUNT, KEY, AND DATA


Clll1l111.lnd Code 1 E (hex) single track
9E (hex) multi-track

(Figure 2-19)

mand Retry is not used or is unsuccessful when an


overrun or check condition occurs, Channel End, Device
End, and Unit Check are presented to the channel.

Transfers the next record encountered on the track from


tlw disc to main storage (excluding Record Zero).
Correction codes following each area are used to check
for data validity. A parity bit is added to each byte prior to
i.ts transfer to the channel.

DATA ADDRESS-Specifies main storage location where


first byte of count data is to be transferred.

Channel End and Device End are presented to the


channel upon completion of the correction code of the
areas.

FLAGS-Used at discretion of the programmer.


COUNT-Specifies the number of key and data area
bytes to be read.

A data overrun or check condition, if detected, initiates


an SCU recovery attempt by Command Retry. If Com-

NDEX
~
~

INITIAL STATUS-Normally zero.

TRANSFERRED ON COMMAND

r-

Gl[]OME G21 COUNT IG2[JEY G2EJATA G31 COUNT IG2[JEY G21 DATA IG3

D~~A G4~NDEX
~

RECORD

I.---

k=

RO F I E L D - - - J

RI F I E L D d
(AND ALL OTHERS FOLLOWING)

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

~~~T

II I I I I I I I I
F

~::r-T

KL

DL

DL

EC

II"L..

2BYTE HD ADRS
KEY LENGTH
RECORD

2-29

307. 302

307.302

045 (04D)

306

HI" \ll; OR 81;

RlAD DATA

If challwd from a count flUid operation,


wad data f,,,ld of the same record. If
,,"allwd from a data field operation,
,-"ad Ih" data field of the next record.

SEND CHNL END


AND DEVICE END
TO CHNL

If "hamed from a Count Field coum


mand operation, read key and data of
__
same record_ If chained from a data
field operation, read the nexl record.

READ KEY
AND DATA

Hf AD AND CHECK
U:C: fOR KEY FLD

044 (041)
IF NOT ORIENTED
TO RN FIELD.
CLOCK OVER
APPROPRIATE
FIELDS

NOTE

NOTE

ThiS command transfers the next data


finlel read on the drive to the system.
(RO data is read only if RO count or Key
is Ihe orientation.)

ThiS command transfers the next key


and daW field (except RO) read on the
seleGled drive to the system. (RO key is
warl only if RO count is the orientalIOn.)

306

044 (05F)
LOAD TAGS TO
RD DATA FIELD
AND GIVE CORRECT
XFER LGTH TO
CONTROLLER

IF NOT ORIENTED
TO RN FIELD,
CLOCK OVER
APPROPRIATE
FIELDS

044 (04D)
045 (05F)
READ DATA FLD
AND XFER DATA
TO CHNL

LOAD TAGS TO
RD KEY FIELD
AND GIVE CORRECT
XFER LGTH TO
CONTROLLER

045 (04D)
READ KEY FLD
AND XFER DATA
TO CHNL

READ AND CHECK


ECC FOR THE
DATA FLD

ISEE FIGURE 2-38)

Figure 2-19. Read Data, Read Key and Data, and Read Count. Key, and Data Flow Diagrams (1 of 2)

3672 .21 -0001-10175

2-30

307 302

NOTE
This command transfers the count, key.
and data fields of the next field (except
ROlon the track

READ COUNT.
KEY. AND
DATA

HEX 1E OR BE

042 (041)
IF NOT DATA
ORIENTED.
ORIENT TO
NEXT DATA FLO
042 (05F)
LOAD TAGS TO
RD COUNT FIELD
AND GIVE CORR
XFER LGTH TO
CONTROLLER
042 (040)
READ AND XFER
eNT AREAS TO
CHNl: CC. HH.
R. KL. AND DL
042 (040)

SAVE KL AND DL
BYTES

042 (040)

READ AND CHECK


ECC FOR CNT FLO

YES

071

Figure 2-19. Read Data. Read Key and Data. and Read Count. Key. and Data Flow Diagrams (2 of 2)

3672.21-0001-10/75

2-31

READ RECORD ZERO

(Figure 2-20)

overrun or check condition occurs. Channel End, Device


End. and Unit Check are presented to the channel.
(Command Retry is not used if a correctable data error;
i.e .. error burst of 11 bytes or less. is detected in the data
area.)

Command Code: 16 (hex) single track


96 (hex) multi-track
Transfers count, key, and data areas of Record Zero (RO)
from the disc to main storage. Data transfer of the RO
count area is initiated by the SCU, which searches for
index and reads gap 1, home address, and gap 2.

DATA ADDRESS-Specifies main storage location where


first byte of RO count data is to be transferred.
FLAGS-Used at discretion of the programmer.

Correction codes following each area are used to check


for data validity. A parity check bit is added to each byte
prior to its transfer to the channel.

COUNT -Specifies number of count, key, and data bytes


to be read.

Channel End and Device End are presented to the


channel upon completion of the correction code check of
the areas.

INITIAL STATUS-Normally zero.

(Figure 2-20)

A data overrun or check condition, if detected, initiates


an SCU recovery attempt by Command Retry. If Command Retry is not used or is unsuccessful when an
overrun or check condition occurs. Channel End, Device
End. and Unit Check are presented to the channel.

Transfers the eight bytes (CC, HH. R. KL, and DL) of the
next count area encountered on the track (excluding RO)
from the disc drive to main storage.

DATA ADDRESS-Specifies main storage location where


first byte of count data is to be transferred.

Correction codes following each count area are used to


check for data validity. A parity check bit is added to each
byte prior to its transfer to the channel.

FLAGS-Used at discretion of the programmer.

Channel End and Device End are presented to the


channel upon completion of the correction code check of
the areas.

COUNT-Eight.
INITIAL STATUS-Normally zero.

SPECIAL
CHARACTERISTICS-Command
execution
begins immediately if Read RO is chained from a Search
Home Address or Read Home Address. In these cases,
the SCU will not search for index.

A data overrun or .check condition. if detected. initiates


an SCU recovery attempt by Command Retry. If Command Retry is not used or is unsuccessful, when an

G1. G2. AND G3 FIXED LENGTH GAPS

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

G4 VARIABLE LENGTH GAP

~~UE~T

READ COUNT
Command Code: 12 (hex) single track
92 (hex) multi-track

I I I I I II I I I
r

FL T

--r
I

2-BYTE CYL ADRS

DL

DL

EC

T ' B J E ERR CORR


DATA LENGTH

2-BYTE HD ADRS
KEY LENGTH
RECORD

~~UE~T

II I I I II I II I
F

FL T

--r
I

KL

DL

DL

EC

I I

T - s J E ERR CORR
2-SYTE CYL ADRS
DATA LENGTH
2-BYTE HD ADRS
KEY LENGTH
RECORD

2-32

READ HOME ADDRESS

(Figure 2-20)

Command Code: 1A (hex) single track


9A (hex) multi-track
Transfers the F. CC. and HH bytes of the home address to
main storage
Correction codes following each home address area are
used to check for data validity_ A parity check bit IS added
to each byte prior to Its transfer to the r.hannel
Channel End and Device End are presented to the
channel upon completion ot the correction code check ot
the areas

A data overrun or check condition. if detected. initiates


an SCU recovery attempt by Command Retry If Command Retry is not used or is unsuccessful when an
overrun or check condition occurs. Channel End. Device
End. and Unit Check are presented to the channel.
DATA ADDRESS-Specifies main storage location where
first byte of home address is to be stored
f.LAGS--Used at discretion of the programmer
COUNT --Five
INITIAL STATUS --Normally zero

Gl G2 AND G3, FIXED LENGTH GAPS


G4, VARIABLE LENGTH GAP

2-33

042 (040)

READ AND CHECK


7 ECC BYTES

NOTE
YES

This command transfers the contents of


the next count field (CC, HH, R, KL,
and DL) to the system. RO count field
is ignored.

HEX 12 OR 92

042 (041)
NOTE
This command transfers the contents of
RO (count, key, and data) field to the
system.

READ
RECORD ZERO

HEX 16 OR 96

IF NOT DATA
ORIENTED.
ORIENT TO
FIRST DATA
FIELD

LOAD TAGS TO RD
KEY FIELD AND
GIVE CORRECT
XFER LGTH TO
CONTROLLER
045 (040)

042 (05F)

READ KEY FIELD


AND TRANSFER
TO CHANNEL

044 (040)

YES
045 (040)

READ AND CHECK


7 ECC BYTES

READ AND CHECK


7 ECC BYTES

ORIENT TO
INDX AND
CLOCK HA

LOAD TAG TO
RD COUNT
FIELD AND GIVE
CORRECT XFER
LGTH TO
CONTROLLER
042 (040)
READ AND XFER
CNT AREAS TO
CHNL: CC. HH,
R. KL, AND DL

YES

042 (040)

YES

READ AND CHECK


7 ECC BYTES
042 (05F)
LOAD TAGS TO RD
COUNT FIELD AND
GIVE CORRECT
XFER LGTH TO
CONTROLLER
042 (040)
READ AND XFER
CNT AREAS TO
CHNL: CC, HH,
R. KL, AND
DL BYTES

LOAD TAGS TO
RD DATA FIELD
AND GIVE
CORRECT XFER
LGTH TO
CONTROLLER

306

SEND CHNL END


AND DEVICE END
TO CHNL

NO

(SEE FIGURE 238)

SEND CHNL END


AND DEVICE END
TOCHNL

044 (040)

READ AND XFER


DATA TO CHNL

071
306
(SEE FIGURE 2-38)

306

Figure 2-20. Read Record Zero, Read Count. and Read Home Address Flow Diagram (1 of 2)

3672 .21 -0001-1 0175

2-34

NOTE
HEX lA OR 9A

READ
HOME ADDRESS

This command transfers five bytes of


track address information (f'CCHH)
from the selected drive to the system.

041 (04A)

GET INDEX
ORIENTATION

041 (040)

READ AND CHECK


7 ECC BYTES

041 (05F)

LOAD TAGS TO RD
HA FIELD AND
GIVE CORRECT
XFER LGTH TO
CONTROLLER

041 (040)

RD HA FLO FROM
CONTROLLER AND
XFER TO CHNL:
FLAG. CC. HH

071
SEND CHNL END
AND DEVICE END
TO CHNL
(SEE FIGURE 2-38)

306

Figure 2-20. Read Record Zero. Read Count. and Read Home Address Flow Diagram (2 of 2)

2-35

3672.21-0001-10/75

READ IPL
Command Code 02 (hex)

(Figure 2-21)

Performs an initial program load (IPL) by causing the SCU


to seek to cylinder 0 and track 0 of a selected drive, and
search for index. Upon detecting index, the R1 data area
is read. This command is normally initiated by setting the
storage device address in the LOAD ADDRESS switches
on the FE panel and pressing the IMPL switch.
Correction codes following each data area are presented
to the channel used to check for data validity. A parity
check bit is added to each byte prior to its transfer to the
channel.
Channel End and Device End are presented to the
channel upon completion of the correction code check of
the areas.

A data overrun or check condition, if detected, initiates


an SCU recovery attempt by Command Retry. If Command Retry is not used or is unsuccessful when an
overrun or check condition ocurs, Channel End, Device
End, and Unit Check are presented to the channel
(Command Retry is not used if a correctable data error;
i.e., error burst of 11 bytes or less, is detected in the area
data area.)
DATA ADDRESS-Specifies main storage location where
first byte of data is to be transferred.
FLAGS-Used at discretion of the programmer.

READ SECTOR
Command Code 22 (hex)

(Figure 2-21)

Execution of this command resets orientation information in the SCU.

Transfers one byte of data from the SCU to main storage.


The byte transferred will normally contain the sector
number (0-127 dec) required to access the last record
processed. If a drive Power On sequence or system reset
occurred after a record was processed, this byte will be
zero. If the last record processed was an overflow record,
this sector number is that of the last segment.
Otherwise, the value will be that of the last record
processed or the last set sector value loaded. for that
drive.

Channel End and Device End are presented to the


channel after the sector number has been transferred.
DATA ADDRESS-Specifies the main storage location
where the sector number is to be stored.
FLAGS-Used at discretion of the programmer.
COUNT-One.
INITIAL STATUS-Normally zero.

COUNT-Specifies number of bytes to be transferred.


CHAINING REQUIREMENT-Must not be preceded by a
Set File Mask in the same chain.

ONE BYTE OF OATA TRANSFERREO ON COM


MAND. THIS TEUS THE USING SYSTEM WHERE
THE START OF THE LAST RECORO OPERATED ON
IS LOCATED.

~'-------------v'----------~

NDEX Gl~OME
~
~

G21 COUNT IG2BEY G2BATA G31 COUNT IG2BEY G2BATA G3

D~~A G4~NDEX
~

RECORD

I - - - R O FIELD----l

~RI FIELD~
(AND ALL OTHERS FOLLOWING)

Gl

G2. AND G3 FIXED LENGTH GAPS


G4 VARIABLE LENGTH GAP

Gl. G2. AND G3 FIXED LENGTH GAPS


G4, VARIABLE LENGTH GAP

2-36

Address of SCU and Dnve are


up on uSIng systm console

This command transfers one byte of


information to the using system. This
tells the using system where the start
of the record last operated on is
located.

s~1

044 (041)

OPERATIONAL OUT drops for 6/lsec.

ALL REG'S ARE


RST BY PWR ON
RST CMD

ORIENT TO Rl
KEY FIELD
(HA AND RO
FIELDS ARE
SKIPPED AND Rl
IS READ)

START 1/0
FORCED IN CPU

LOAD TAGS TO RD
DATA FIELD AND
GIVE CORRECT
XFER LGTH TO
CONTROLLER

044 (OSF)

=:]---

304

READ SECTOR

048 (03B)
READ SCTR NO.
FROM THE DRV
TO THE SCU

044 (040)
READ DATA AND
XFER DATA FLO
TO CHNL

INITIAL
SELECTION
Th,s command causes the data field of
R1 on track zero, head zero of the
selected Drive, to be transferred by the
SCU to the uSing system. This data is
ThIS command IS normally set up by
IPL button. but may be Start I/O
InstructIon. If in a chain It must not be
preceded by a Set File Mask command,
or command IS rejected.

...-_ _--1._ _.;..30;...,4

048

044 (040)

~~~:::~I~t~~:~!~Ou;~~~ ;~~~~~a~~f:~
is completed.

- - - j ---

04B
SET SECT NO
TO 0 IF IN 0 OR
1 RANGE. SUBT 1
IF IN LOW SIDE
OF SECTOR

READ AND CHECK


ECC BYTES

READ IPL

MASK OFF ZERO


BIT AND SEND
MODIFIED VALUE
TO CHNL

306

047 i03E)
YES
Control Unit issues recalibrate to dnve
Instead of seek

SEEK TO CYL 00
AND HEAD 00

044 (04F)

PRESENT CHNL
END AND DEVICE
END TO THE CHNL

RESTART OPER
(NO ERR CORR
PROC IN USING
SYSTEM)

306

Figure 2-21 Read IPL and Read Sector Flow Diagrams

2-37
3672 .21 -0001- 10/75

2.6.3 Read Data Transfer

2.6.4.2 SYSTEM

The read data transfer function is described by means of


the flow diagram and block diagram in Figure 2-22 and
the microprogram subroutine in Figure 2-23. The
flowchart is keyed to the block diagram and
microprogram subroutine. Conventions for reading the
microprogram subroutine are explained in Section 4.

Executes a Start //0 instruction which addresses a


diagnostic channel command.

After receiving Device End status. executes a Start


//0 instruction which address a Read Diagnostic
Status 1 command.

2.6.4.5 650 DRIVE


Provides drive to disc.

Provides for movement of head from track to track.

Picks up data Signals from disc via read head.

2.6.4.6 650 INTERFACE

2.6.4 Read Diagnostic to System


2.6.4.1 INTRODUCTION

2.6.4.3 CHANNEL

Read diagnostics to the system are used to transfer


diagnostic tests stored on the 650 Flexible Disc to the
CPU of the using system. Any diagnostic test stored on
the 650 Flexible Disc can be selected by the u'sing
system. To transfer the diagnostic test. the using system
must execute a Diagnostic load followed by a Read
Diagnostic Status 1 command. The Diagnostic Load
command will transfer 668 bytes (containing the specific
diagnostic) from the 650 Flexible Disc to control storage.
The Read Diagnostic Status 1 command then transfers
the 668 bytes from control storage to the using system
main storage.

Transfers address and command information to the


selected subsystem.

Transfers status of subsystem to the system.

Transfers
storage.

Figure 2-24 illustrates a Read Diagnostic to System


operation. The operation is described in the following
paragraphs.

diagnostic

from

subsystem

to

system

Controls timing of data to and from the channel.

Checks parity of data transferred to the SCU.

Decodes the Diagnostic Load command.

Sends Channel End status to the channel.

Sends start. seek. and read control lines to the 650


interface according to the control byte sent with the
Diagnostic Load command.

Starts 650 drive motor.

Moves head to track selected by the number of


Diagnostic Load commands.

Transfers diagnostic data from the 650 Flexible Disc


to the control storage.

Reads 668 bytes of data from the disc.

Transfers Device End to the channel after data is read


into control storage.

2.6.4.7 CONTROL STORAGE

Decodes Read Diagnostic Status 1 command.

Stores microprogram to control the operation of


subsystem.

Controls transfer of data from control storage to the


channel.

Stores diagnostic data read from 650 Flexible Disc on


diagnostic load operation for transfer on the Read
Diagnostic Status 1 command.

Transfers subsystem status to the channel.

Controls transfer of data (668 bytes) from control


storage to the channel.

Sends disc drive and SCU status to the channel.

2.6.4.4 CHANNEL CONTROLS

2.6.4.8 CONTROL UNIT

Reads diagnostic data out to SCU via a Diagnostic


Sense operation.

2-38

~
I

SELECTED TO
CHANNEL.
LOCK CHANNEL
SWITCH IS ON
CHC BIT 4

CONTROLLER
RAISES TAG
VALID, INDICAT
ING THAT IT WILL
PERFORM THE MINIORDER

DATA BYTES
STORED IN R03
REG. AS THEY
ARE RECEIVED
FROM THE CONTROLLER

RAISE OPERATE
UP AND
UNSQUELCH
READ HEAD TO
CONTROLLER

ORDER SENT

TO CONTROLLER
(READ-Gl, ETC.)
CO=/SE
00=/47

I
ALL BYTES READ
WHEN DATA
TRANSFER
COUNTER DECRE
MENTS TO 0 &
INTERRUPT CONDITION OCCURS

I
DROP CHANNEL
TRANSFER AND
CHECK FOR
ERRORS FROM
DRIVE CHC BIT

DATA BYTES SENT


TO CHNL VIA
BI REG IF THERE
WAS A CHNL
BUFFER READY

FIRST SYNC IN
RECEIVED FROM
CONTROLLER.

REFER TO READ
LOOP SUBROUTINE,
FIGURE 2-23

CHANNEL TRANSFER STARTED.


SCU TRANSFERS
NUMBER OF
BYTES INDICATED BY DTC
SETIING

IF NO CHNL
BUFFER READY
CHNL TRUNCATION INDICATED IN
FIELD CONTROL
BYTE

PRESENT ENDING
STATUS

INITIAL STATUS
SENT TO
CHANNEL
CT=/14
BI=IOO
(STATUS)

~ CO ~

CONTROLLER

~ DO~ IT;;;;;Sf~ ;';;;t~s ~s~u~h;:;ri-;l

CONTROL AND
STATUS DATA

I~---=---------,

Transfer bytes read from BI Register to


a buffer register into the channel.
Maintains sync with channel.

TO SYSTEM

CHANNEL

Controls selection of module and read


circuits. Checks number of bytes read
to end operation.

I
I

circuits to read data.


CUDI

rn
I

CONTROL AND

STATUS DATA

~:t~dt~S t::'~~~~D~~~ read circuits send

DRIVE

SCU

. . - DI

Figure 2-22 Read Data Transfer Flow and Block Diagram

2-39

~----8DE

HA.CNT.KEY --8DC

------l

CNT=/E8
CIBUF=R03.INC)

Bl-073-2-- 800
STORE 4 BYTES IN
BUFFER LOCATION
FROM RTN O4C

PROCESS.RO

.------8EO

-BIDBR+INTi
+"0
B4-081l--8DF
WAIT FOR & THEN
SET RO '0 ~ FIRST
DATA BYTE

BIINT)
NO.OBR
B5-OB4-1--921
BR IF NO DBR.

81=R3
-BICBR)
TRUNC.B.3
B3-078-2--8EB
XFER BYTE TO CHNL
BR IF NO CBR

ALL. DONE-- 8F9


PS=B.G4

TRUNC. B. 3 - - 8EB

CLOCK. RO--8FO

.------8Fl

BYTE.O - - - 8 E l
}--+--+-+---'~BI RO

-BICBR)
TRUNC.B.C
F2-056-1--8EC
SEND DATA 8YTE TO
CYNL BR IF NO CBR

CHCo 08

RC=DI

BIBR5)
CLOCK.RO
D3-ll0-1--BFO
SET TRUNCATED',
BR IF CHNL INT

B(lNT)
ALL. DONE
D4-l20-2--BF9
DROP CHNL XFER.BR
WHEN DTC=O

BIUNCOND)
BYTE.O
D5-122-1--8El
DATA BYTE INTO
ROiO

PROCESS.Rl-8El

.------8E3

RI=DI

CNT E8

-BIDBR+INT)

B(lNT)
NO.DBR
F4-091-2-- 921
BR IF NO D8R

-./0
F3-088-1--8E2
WAIT FOR & THEN
SET Rl ,0 - NEXT
, DATA BYTE

BIBR1)
SC=O.WAIT
C7--138-2-- 936
BR IF DTC = O.
INORMAL EXIT)

G 58FA
=:J------DA
BIADR=SR)
C8-140-2
SET 'NO D8R' ERROR
RETURN ...

04006 B1

TRUNC. B. 0--8EC
0400602

B18R5)
CLOCK.Rl
H3-ll2-l--8F2
SET 'TRUNCATED
BR IF CHNL INT

r-----8F3
RI=DI
8IUNCOND)
BYTE. 1
J5-126-1--BE4
DATA BYTE INTO
R1 0

Figure 2-23, Read Loop Microprogram Subroutine

2-40

FLEXIBLE DISC DRIVE

DIAGNOSTIC
DATA

FLEXIBLE DISC DRIVE INTERFACE

DIAGNOSTIC
DATA

ADDRESS DATA
COMMAND DATA

--

SYSTEM

STATUS AND
DIAGNOSTIC DATA

CHANNEL

CHANNEL
CONTROLS

SERVO CONTROL
READC ONTROL
POWER ON CONTROL

STORAGE
CONTROL UNIT
(SCU)

DIAGNOSTIC
DATA IN

ADDRESS
CONTROL

SERVO DRIVE

fDIAGN OSTIC SENSE DATA


OUT. AN D DIAGNOSTIC
MICROP ROGRAM

CONTROL STORAGE

Figure 2-24 Read Diagnostic to System Operation

2-41

2.7 WRITE COMMANDS

2.7.1.3 CHANNEL INTERFACE

Controls timing of transfer of data between the


channel and the SCU,

Checks parity of data transferred to the SCU,

Transfers the address where data is to be stored to


the SCU.

Transfers data from the using system to the SCU.

Transfers ending status to the channel.

Changes data pulses from SERDES to currents for the


write head.

Indicates retry status to the channel if the error is a


type that can be retried, and sets up to relocate the
record,

Provides the SCU with status of the drive,

Index pulse generated from servo data and sent to


SCU.

2.7.1 Operation Block Diagram


Write commands store data from the using system on the
disc pack of a selected drive. Format write commands

Write Home Address, Write RO; Write Count, Key, and


Data; and Erase cause the balance of the track to be
changed. The Write Key Data and Write Data commands
change only the key and/or data fields of a record to be
changed (field length stays the same).

2.7.1.5 CONTROLLER

Transfers write control information to the selected


drive.

Transfers the status of the selected drive to the


Microprocessor.

2.7.1.4 MICROPROCESSOR
The following text is keyed to the Write Operation block
diagram shown in Figure 2-25.

2.7.1.7 READ/WRITE HEAD

Changes currents in write windings to magnetic flux.

Selected head writes data.

PLO head provides PLO pulses.

Servo head and circuits hold access at selected track.

Decodes the write command.

Controls gap configurations.

Selects the addressed drive and head.

Receives data serial-by-byte from the Microprocessor.

2.7.1.1 SYSTEM

Checks the status of the drive and transfers the status


to the channel.

Changes data byte to serial-by-bit data.

The system executes the Start //0 instruction which


addresses a channel command word (CCW) containing a
write command,

Sets up the write controls in the controller.

Sends serial data to ECC.

'2.7.1.8 PACK

Controls the byte-by-byte transfer of write data from


the channel interface to the controller.

Sends serial data to the MFM for precompensation.

The pack receives and stores bits of information on the


magnetic coating of the disc.
/

Uses VFO trigger pulses to control write timing,

Checks subsystem for errors,

2.7.1.2 CHANNEL
The channel executes the CCW to transfer the write
command to the SCU.

l~sl-m,

SYSTEM -

;';;-IT-;- -

-l

~~6A

Monitors incoming tags for error indicators.

IIF

1 WRITE

BUS OUT

CTL
IIF

WRITE
DATA

DATA

Each write command is described in the following pages


by means of a flow diagram and a narrative description.

Sets up the drive to write on the disc pack.

CHNL I s C U - - - - - - - - - - - - - - - - 1

CONTROL
- - - - - - - - - . INFO

2.7.2 Command Descriptions


2.7.1.6 WRITE CIRCUITS

,r----'

l~c~I~---------------1

WRITE
DATA
(SERIAL)

I
I

WRITE
DATA

WRITE
DATA

PLO

BUS IN
CONTROLLER
SYSTEM

CHANNEL

CHAN~,EL

DEVICE
STATUS
INFO

INTERFACE

DEVICE
STATUS
INFO

I
I1_ _ _ _ _ _ _ _ _ _ _ _

L __________ _

TAG OUT

INDEX
WRITE
CONTROL

TAG IN

STATUS

MICROPROCESSOR

_____ J

L ___ ..JI

WRITE
CIRCUITS

SERVO
TRACK
DATA

READ/WRITE
HEAD

SERVO
TRACK
DATA

I1_ _ _ _ _ _ - - - - - - - - -

PACK

____ _

2-42

WRITE DATA
Command Code 05 (hex)

(Figure 2-26)

FLAGS-Used at discretion of programmer.


COUNT-Specifies number of data bytes to be written. If
CCW Count field is less than the formatted record DL, the
SCU writes zeros in the remaining data area, followed by
writing ECC bytes, and presents Channel End and Device
End to the channel. If CCW count field is greater than the
formatted record DL, the SCU writes only the number of
bytes indicated in the DL and then writes ECC bytes.

Performs normal record updating after track formatting.


Execution of command causes specified data in main
storage to be written in data area of selected record.
Number of bytes written is specified in the CCW count
field. It may be less than the data length (OL) specified in
the formatted record.

CHAINING REQUIREMENTS-Must be chained from


Search 10 or Search Key Equal. (The Search command
must compare equal on all bytes of the searched field.) If
chaining requirement is not met Unit Check is presented
in initial status.

Channel End and Device End are presented after the ECC
bytes have been written for the data area.
DATA ADDRESS-Specifies main storage location of
data used to update record.

WRITE KEY AND DATA


Command Code 00 (hex)

(Figure 2-26)

COUNT-Specifies number of key and data bytes to be


written. If CCW count is less than the formatted record
KL/DL, the SCU writes zeros in the remaining areas,
followed by writing ECC bytes, and presents Channel End
and Device End to the channel. If CCW count is greater
than the formatted record KL/DL, the SCU writes only
the number of bytes indicated in the KL/OL and then
writes ECC bytes.

Performs record updating after track formatting. Execution of command causes specified data in main storage to
be written in key and data areas of selected record.
Number of bytes written is specified in the CCW count
field. It may be less than the key length (KL) and data
length (DL) specified in the formatted record.

INITIAL STATUS-Normally zero.


Channel End and Device End are presented after the ECC
bytes have been written for the data area.
CHAINING REQUIREMENT-Must be chained from a
Search 10 Equal command. (The Search 10 command
must compare equal on all bytes of the searched field.) If
chaining requirement is not met. Unit Check is presented
in initial status.

DATA ADDRESS-Specifies main storage location of


data used to update record.
FLAGS-Used at discretion of programmer.

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

~~E~T

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

I I I I I I I I I I
F

KL

T --rI. I

FL
2BYTE CYL ADRS
2BYTE HD ADRS

DL

DL

EC

T , ' B J E ERR CORR


DATA LENGTH
KEY LENGTH

RECORD

~~E~T

I I I I I I I I I I
F

T --rI

FL
2BYTE CYL ADRS

KL

I I

DL

DL

EC

T , ' B J E ERR CORR


DATA LENGTH

2BYTE HD ADRS
KEY LENGTH
RECORD

2-43

WRITE COUNT, KEY, AND DATA


Command Code 1 D (hex)

(Figure 2-26)

FLAGS-Used at discretion of the programmer.


COUNT-Specifies number of count, key, and data bytes
(8 + KL + DL) to be written. If CCW count area is less than
8 + KL + DL, the SCU writes zeros in the remainder of the
record.

Causes the count, key, and data area of a record in main


storage to be written on a selected drive. The count area
is made up of the first eight bytes from main storage. The
flag byte is generated by the SCU; the remaining data is
written in the key and data areas as specified by the KL
and DL bytes in the count area.

INITIAL STATUS-NormallY zero.


CHAINING REQUIREMENTS-Must be chained from
Write Record Zero; Write Count, Key, and Data; Search
10 Equal, or Search Key Equal CCW. If chaining requirement is not met, Unit Check is presented in initial status.

Channel End and Device End are presented to the


channel after correction code bytes are written for the
data area.

A Read Data or Read Key and Data command may be


inserted between Search command and Write Count,
Key, and Data command.

DATA ADDRESS-Specifies main storage location where


count, key, and data bytes of record are located.

TRANSFERRED ON COMMAN0-1

r::l
r::l 0 0 0 BOD
~G1L:JG~c:rl~:JG2CJG3
COUNT

R~i~D
DATA

G2LJG2CJG3

r::l
G4'V

~RIFIELDd

/..--ROFIELD----l

(AND ALL OTHERS FOLLOWING)

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

~~~T

I I I I I I I I I I I
T ---rI I I
F

KL

DL

DL

EC

FL
T , ' B J E ERR CORR
2BYTE CYL ADRS
DATA LENGTH
2BYTE HD ADRS
KEY LENGTH
RECORD

244

FROM
CMD 7.11
Chained from a Search Equal I D com
mand.

FROM
(2 OF 21

FROM
FIGURE 229

306

~-

WRITE KEY

--

L-_A_N_D"""'T"DA_T_A_~

This. command updates the key and


data field of a record selected by the
using system. This command must be
chained from a successful Search Equal
I D command on the COunt field of the
same record. If key length is zero,
Write Key Data is the same as Write
Data.

READ KEY BYTES


SENT FROM
CHANNEL AND
GIVE TO
CONTROLLER

Write number of
bytes given in count

field key leng!n byte.

SET CHNL END


AND DEVICE END
IN BUS IN AND
RAISE STATUS
IN TO CHNL

062 (06D)
RECEIVE RESPONSE
FROM CONTROLLER
TO SHOW
COMPLETION

YES

062 (05F)
Write 1 Byte SCU+OEV
address and 46
bytes of zeros

LOAD TAGS TO
WR G2 AND GIVE
CORR XFER LGTH
TO CONTROLLER
FO DATA FIELD

CONTROLLER
CONTINUES
TO WR ZEROS IN
G4; SCU BUSY
TO ALL CHNL I/Fs

302

062 (06D)
TO
(2 OF 2)

If data length bytes


.n the count field
are zero (EOF). write
one byte of zero

READ DATA BYTES


SENT FROM CHNL
AND GIVE TO
CONTROLLER

Write number of bytes


given in count

field data length bytes.

Chained from a Search Equal Key com-

mand.

062 (06D)

This command updates the data field of


a record selected by the using system.
This command must be chained from a
successful Search Equal I D or Search
Equal Key command on the same
record.

L-_W_R_IT. ,E~D_A_T_A_

... --

RECEIVE RESPONSE
FROM CONTROLLER
TO'SHOW
COMPLETION

BRING DOWN
OPERATE TAG

306

Figure 2-26. Write Count, Key, and Data; Write Key and Data; and Write Data Flow Diagrams (1 of 2)

3672.21.-0001-10/75

2-45

FROM
FIGURES
2-27.2-28.2-30

SEND ID. CAR.


HAR. AND FLAG
TO CONTROLLER
FROM BUFFER

062 (060)
READ CC. HH. R#.
KL. AND DL FROM
CHNL AND SEND
TO CONTROLLER
062 (06D)
Chained from a Write RO; Write Count.
Key. and Data; Erase; or a.successful
Search Equal 10 or Search Equal Key
command

FROM
(1 OF 2)
A

RECEIVE RESPONSE
FROM CONTROLLER
TO SHOW
COMPLETION

_ _ _ _H_E_X_ID_I- -

-;

WRITE COUNT.
KEY. AND DATA

---r-_...

......

I
The Write Count. Key. and Data command writes a record (count. key. and
data) on the disc and track selected
by the system. Write CKD must be
chained from a Write RD. successful
Search Equal ID. or Search Equal Key
command

I
I

_ _ _ ...J

LD TAGS TO
WR G2 AND GIVE
CORR XFER LGTH
TO CONTROLLER
FOR RN KEY FIELD

LD TAGS TO
WR G3 AND GIVE
CORR XFER LGTH
TO CONTROLLER
FOR CNT FIELD

TO

TO

(1 OF 2)

(1 OF 2)

Figure 2-26. Write Count. Key. and Data; Write Key and Data; and Write Data Flow Diagrams (2 of 2)

3672.21 -0001-1.0/75

2-46

WRITE SPECIAL COUNT, KEY,


AND DATA

(Figure 2-27)

FLAGS-Used at discretion of the programmer.

Command Code 01 (hex)

COUNT-Specifies number of bytes in the count, key,


and data areas bytes (8 + KL + DL) to be transferred. If the
CCW count is less than 8 + KL + DL, the SCU writes zeros
in the remainder of the record.

Formats a segment of an overflow record, the last


segment of which is written by a normal Write Count.
Key, and Data command. The count area of the record is
made up of the first eight bytes from main storage. The
flag byte contains a one written in bit position 4. This bit,
written by the SCU, indicates that another part of the
next record is located on the next track. Correction code
bytes are written by the SCU at the end of each record
area.

INITIAL STATUS-Normally zero.


CHAINING REQUIREMENTS-Must be chained from a
Write Count, Key, and Data; Search 10 Equal; or Search
Key Equal command. (The search commands must
compare equal on all bytes of the searched field.) A Read
Data or Read Key and Data command may be inserted
between the Search command and Write Special Count.
Key, and Data command. If chaining requirements are
not met, Unit Check is presented in initial status. An
overflow record must be the last record on the track. A
subsequent overflow record must be the only record on
the track (excluding RO).

Channel End and Device End are presented to the


channel after correction bytes have been written for the
data area.
DATA ADDRESS-Specifies the main storage location
where the count, key, and data areas of the record are
located.

Chained from a Wrote RD. Write


CK-O. or a successful Search
Equal 10 or Search Equal Key
command.

=:]
HEX 01

WRITE SPECIAL

- - - - - - __
_ C_O_U_N_T.,'_K_E_Y_'
ANO DATA

NO

TRANSFERRED ON COMMAND

~ml~~I~I~B~~B~~~I~B~~B~~~~~~
"
"
~
~

~
RECORD

I.----ROFIELO-..l

SET OVERFLOW
BIT IN BUFFER
TO WRITE
OVERFL9W REC

~RIFIELD~
lAND ALL OTHERS FOLLOWING)

TO
. FIGURE

Gl. G2, AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

FL

~
I I I

2-26

T ' B J E ERR CORR

2BYTE CYL ADRS


DATA LENGTH
2BYTE HD ADRS
KEY LENGTH
RECORD

Figure 2-27. Write Special Count, Key, and Data Flow Diagram

3672.21-0001-10/75

2-47

WRITE RECORD ZERO


Command Code 15 (hex)

(Figure 2-28)

Causes specified data in record zero (RO) count, key, and


data bytes in main storage to be written on selected
drive. The count area is made up of the first eight bytes
from main storage. The flag is generated by the sCU; the
remaining data is written in the key and data areas
specified by the KL and DL bytes in the count area.
Channel End and Device End are presented to the
channel after correction code bytes are written for the
data area.
Proper operation with Operation System (05) requires an
eight-byte data field in RO.

DATA ADDRESS-Specifies main storage location of RO


count, key, and data bytes.
FLAGS-Used at discretion of the programmer.
COUNT-Specifies number of count, key, and data bytes
(8 + KL + DL) to be written. If CCW count area is less than
8 + KL + DL, the sCU writes zeros in the remainder of the
record.
INITIAL STATUS-Normally zero.
CHAINING REQUIREMENT-Must be chained from a
successful Write Home Address, or Search Home Address Equal command.

=:J----

DECODE
COMMAND
AS WRITE
RECORD ZERO

This command writes the RO of a track.


The record can be a count-key-data
format record. If written under Operation System (OS) control, RO is used to
describe the track. If the track is defective, the 10 of the alternate track is
written in the. count field of RO. On
normal records, the RO data field gives
the number of records written on this
track and the number of bytes remaining on the track. RO must be chained
from a Write Home Address or a
successful Search Home Address command.

NO

304
SET UNIT CK IN
STAT BYTE & CMD
REJ IN SENSE DATA

LOAD TAGS TO
WR G2 AND GIVE
CORR XFER LGTH
TO CONTROLLER
FOR COUNT FIELD

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

TO
FIGURE
2-26

~~E~TIFICIII

IIKLIDLIDLIECI

T -rI I

FL
T , . B J e ERR CORR
2BYTE CYL ADRS
DATA LENGTH
2BYTE HO ADRS
KEY LENGTH
RECORD

Figure 2-28. Write Record Zero Flow Diagram

3672 .21 -0001-10175

2-48

WRITE HOME ADDRESS


Command Code 19 (hex)

(Figure 2-29)

FLAGS-Used at discretion of the programmer


COUNT-Should be five. If less than five, the SCU
records zeros until five bytes have been written. If count
IS greater than five, only the first five bytes at the main
storage location are written.

Establishes track identity, a prerequisite for data


operations on that track. The SCU orients on Index;
writes gap 1, home address, ECC bytes, source ID; and
then writes gap 2 or gap 4. Bit 5 of the home address flag
byte (CE pack) must be zero

DECODE CMD
AS WRT HOME
ADDR

ThIS command wTltes a new Home


Address field on the DTive and track
selected by the uSIng system

INITIAL STATUS-Normally zero.

Channel End and Device End are presented after the ECC
bytes have been written for the data area.

CHAINING REQUIREMENTS-Must be preceded by a Set


File Mask permitting Write Home Address commands. If
requirement is not met. Unit Check is presented in initial
status.

DATA ADDRESS-Specifies main storage location of the


home address bytes (F, CC, and HH).

WAIT FOR
INDEX

--, r
~Gll ~D~E

061 (05F)
TRANSFERRED ON COMMAND

IG21 COUNT I G 2 E J G 2 E J

G3

1 COUNT I G 2 E J G 2 E J G 3

I i '
~ROFIElD

Gl

----l ~THERS

RrFIELD~

G2 AND G3 FiXED LENGTH GAPS


G4 VARIABLE LENGTH GAP

DATA
OF
G4
LAST
RECORD

~
INDEX
PT

LOAD TAGS TO
WR G1 AND GIVE
CORR XFER LGTH
TO CONTROLLER
061 (060'

FOLLOWING

SEND SYNC 10,


CAR. AND HAR
TO CONTROLLER
061 (060)
READ FLAG. CC.
AND HH FROM
CHNL AND SEND
TO CONTROLLER
061 (060)
RECEIVE
RESPONSE FROM
CONTROLLER TO
SHOW COMPLETION

TO
FIGURE 2-26

Figure 2-29. Write Home Address Flow Diagram

3672.21 -0001-10175

2-49

ERASE
Command Code 11 (hex)

(Figure 2-29)

Writes zeros in count, key, and data areas on a selected


drive The address mark is not written with this com
mand The erased record and all records that follow on
the track are unrecoverable.

FLAGS-Used at discretion of the programmer.


COUNT Specifies number of bytes
data areas of the record.

In

count. key. and

CHAINING REQUIREMENTS-

Must be chained from a Write Record


Zero; Write Count, Key and Data
command; Search ID Equal; or Search Key
Equal command. (The Search commands
must compare equal on all bytes of the
searched field.)

Must not be chained from an Erase


command.

Channel End and Device End are presented at the end of


the data area. Remainder of the track is padded with
zeros.
DATA ADDRESS-Specifies main storage location where
count, key, and data areas of the record are located.

Chained from a Write RO, Write


C-K-D, or a successful Search
Equal 10 or Search Equal Key
command.

3----

ERASE

NO

LO TAGS TO ERASE
G3 (WR o's TO INDEX)
AND GIVE CORR
XFER LGTH TO
CONTROLLER

TO
FIGURE
2-26
Gl

G2. AND G3 FIXED LENGTH GAPS


G4 VARIABLE LENGTH GAP

Figure 2-30. Erase Flow Diagram

3672 .21 -0001 -10175

2-50

r-----..,

TURN ON
CHC BITS 0,1
TO PLACE CHNL
INTERFACE
IN WRT MODE

SERVICE OUT /
DATA OUT FROM
CHNL SIGNALS
CHNL INTERFACE
THAT A BYTE
IS ON BUS OUT

/lPROG MOVES BO
TO DO
FOR INPUT TO
CONTROLLER

TURN ON CO 0
TO SEL DRV

CHNL INTERFACE
PLACES BYTE IN
CHNL BUFFER REG

SYNC IN RECEIVED
FROM CONTROLLER,
DO VALIDATED BY
SYNC OUT

SERDES SIGNALS
-; THAT IT IS READY
I WITH B(DBR)

NO

I
I

L _____ .J

YES

NO

IF BO REG IS FREE
BYTE IS XFERD
FROM BUFFER
TO 80 AND
B(CBR) IS
TURNED ON

LOAD TAGS TO WR
FIELD AND GIVE
CORRECT XFER
LGTH TO
CONTROLLER

CHNL INTERFACE
ASKS FOR NEW
BYTE BY RAISING
SERVICE IN:
DATA IN

r----.,

B(CBR) signals the

_ -I microprogram that
a byte IS in BO I
I Register,
I
L _ _ _ _ ...J

SET DBR TO
INDICATE THAT
DO IS READY
FOR NEW LOAD

r-----,

The onterface may

- 1 :~~f:ra a~~ea~:t~:~ I
I on BO Register I
L- _ _ _ _

..J

Figure 2-31. Write Data Transfer (1 of 2)

2-51

2.7.3 Write Data Transfer


The write data function is described by means of the flow
diagram and block diagram in Figure 2-31, and the
microprogram subroutine in Figure 2-32. The flowchart
is keyed to block diagram and microprogram subroutine.
Conventions for reading the microprogram subroutine
are explained in Section 4.

CONTROLLER

,------------,
I
B

CONTROL
AND STATUS
DATA

~~r~~~::r:o c~~i~~OI~a:~. set

up the drive

I
I

CUDI

I
I

FROM SYSTEM
CHANNEL

Transfers bytes from channel bus out


to channel buffer register, then to the
eo register. Maintains sync with the
channel.

I
I
SCU

Controls selection of drive and sends


Operate to Controller. Checks number
of bytes written to end operation.
BYTE BY
BYTE WRITE
DATA

I
I
I

I
I
I

I
I

CONTROL
AND STATUS
DATA

DRIVE

I
SERDES

VFO

SERIAL WRITE
DATA

Head is selected, and write circuits


place data on the disc pack.

I
I

I
I

I
I
I

Transfers full byte from DO register to


buffer
register
and
signals

:;f;:~~~~~~; wi~t~LOe~~~~ t~~~~~


~:!: !~ ~~~c d~;::d. Sends serial write

L _________ --.J

Figure 2-31. Write Data Transfer (2 of 2)

2-52

C C H H - - - - 1Bl

.----_ _ _ _ 1B2

CNT=/BO
C(RESET.DBR)
-B(CBR)
OVER. TRO
2-103-2--1C9
SET LARGE TIMER
BR IF OVERRUN

RO=BO GO

BYTE.0---1B3

B(INT)
DONE-l
B5-106-3--1 BF
READ BUS OUT
OVER. T R O - - 1 C9
GO = 100
C(CH.FREEZE)
-B(lNT)
BYTE.O
C3-157-2--1B3
GO FOR WRrlNG O'S
BR IF NOT DONE
FREEZE CH TRXF

lB6
SOME.ENTRY-1B4
Rl=/oo
-B(CBR)
OVER.TRl
E2-110-2-- lC6
ZERO Rl IC FOR TRN
IS CH READY!

-B(DBR.INT)
*+/0
F6-114-3-- lB6
LOAD DO FOR BUF/W

OVER.TR1-- lC6
GO=/oo
C(CH.FREEZE)
-B(lNT)
BYTE. 1
G3-- 148-2--1 B6
GO FOR WRT ZERO
BR IF NOT DONE

Figure 2-32. Write Loop Microprogram Subroutine

2-53

2.8 SEARCH COMMANDS


Search commands encompass Search Home Address.
Search 10 Equal, Search 10 High Search 10 Equal or
High Search Key Equal, Search Key High and Search
Key Equal or High. Descriptions and flow diagrams for
each of these cO,mmands are presented in the following
pages,
Operation with the Search 10 Equal. Search 10 High and
Search 10 Equal or High commands is as follows:

These commands compare search data from the


using system with data read from the drive.

Only one record can be searched for at a time.

When a search is successful, Status Modifier


(bit 1) is turned on in the status byte with
Channel End and Device End.

If Multi-track (bit 0 of the command byte) is on,


automatic head switching will take place when
Index IS passed

Multi-track (MT) operation


cylinder boundary

SEARCH HOME ADDRESS EQUAL


Command Code 39 (hex) single track
B9 (hex) multi-track

(Figure 2-33)

DATA ADDRESS-Specifies main storage location of a


cylinder number (CC) and head number (HH)
FLAGS-Used at discretion of the programmer.

will

not

cross

When a search is unsuccessful. the Search


command must be reissued by the channel to
continue the search

Transfer In Channel (TIC) command must follow


a Search command to allow continued search,

Passing the index point twice on the same track


or the detection of end of cylinder discontinues
Search operation. Channel End, Device End, and
Unit Check are presented in the ending status
byte, No Record Found is set in sense byte 1 for
single track searches or End of Cylinder for
multi-track operations.

Causes SCU to compare home address (HA) on


designated track with HA read from main storage.
Execution begins with a search for Index. If the single
track command code is designated. the search takes
place on the present track; if the multi-track code is
indicated, the search starts on the next track, When
Index is detected, the cylinder and head numbers from
main storage are compared with those in the track HA If
the comparison is equal. Channel End. Device End. and
Status Modifier are presented to the channel. If the
comparison is unequal, Channel End and Device End
only are presented to the channel.

COUNT-Should be four. If count is greater than four


bytes, the search is completed when four bytes are
received by the SCU and the command is terminated
with Channel End and Device End (and Status Modifier if
the comparison was equal.)
If the count is less than four bytes. a comparison of the
storage and track data continues until the CCW count is
decremented to zero, Channel End and Device End are
presented to the channel when the home address and
correction code bytes are read and checked. Status
Modifier is presented with Channel End and Device End
if the search was satisfied on short field. Chained write is
not allowed by the Controller.

The Flag byte is not transferred to the channel or


compared during command execution, If a Bus-Out parity
error is detected. Channel End, Device End. and Unit
Check Status are presented in ending status,

,SEARCHED WHEN INDEX PT IS DETECTED

NDEX Gl~OME
~
PT

INITIAL STATUS-Normally zero,

ADR

G21 COUNT IG2BEY G2BATA G31 COUNT IG2BEY G2BATA G3

I - - - - - - R O FIELD-----I

D~~A G4~NOEX

LAST
RECORD

PT

~RI FIELD~
lAND ALL OTHERS FOLLOWING)

Gl. G2. AND G3, FIXED LENGTH GAPS


G4, VARIABLE LENGTH GAP

2-54

051
READ C, C, H, H
BYTES FROM
DRIVE & CHNL &
COMPARE

::1--

HEx:l

SEARCH HOME
ADDRESS EQUAL

POST TYPE
OF ERROR

SEND READ GAP 1


ERRORS
ORDER
~------------~
TO CO NT
051 (05F)

SET CHNL END &


DEVICE END
ENDING STATUS

SET CONTROLLER
TO READ MODE

051
051 (04E)
READ ID BYTE
FROM DRIVE

051 (04F)
SET CHNL END,
DEVICE END, &
STATUS MODIFIER
ENDING STATUS

CHECK FOR
MORE ERRORS
DURING SEARCH

SET NO RECORD
FOUND IN
SENSE CONTROL
BLOCK

306

051 (04F)
YES
SET EOC
IN SENSE
CONTROL BLOCK

SET CHNL
INTERFACE
TO READ
MODE
051
READ CAR, HAR,
FLAG BYTES
FROM DRIVE

SET HOME
ADDRESS PREREQUISITE IN
R3/9 TO ALLOW
CtlAINED WRITES

SET CHNL END,


DEVICE END, &
UNIT CHECK .

306
306

Figure 2-33. Search Home Address Equal Flow Diagram

3672.21-0001-10175

2-55

(Figure 2-34)

SEARCH 10 EQUAL
Command Code: 31 (hex) single track
81 (hex) multi-track

detection of end of cylinder, Channel End, Device End,


and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1.
DATA ADDRESS-Specifies main storage address of a
five-byte portion of a count area containing CC, HH, and

Compares an ID in main storage with the next count ID


area on the track (including RO). If an equal comparison
results, Channel End, Device End, and status modifier
are presented to the channel. If an unequal comparison
results, Channel End and Device End only are presented
to the channel.

R.
FLAGS-Used at discretion of the programmer.
COUNT-Should be five. If count is greater than five, only
the first five bytes from main storage are compared.
Channel End and Device End are presented to terminate
the command, and status modifier is presented if the
comparison was equal.

If a single-track search is specified, the search is


confined to one track. The search is repeated until the
search condition is satisfied or until two index points are
detected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record Found is set in sense byte 1.

If count is less than five, a comparison of main storage


and track data continues until the CCW count is zero.
Channel End and Device End are presented to the
channel when the ID and correction code bytes are read
and checked. Status modifier is presented if the search
on the short field is satisfied.

Ifa multi-track search is indicated, the search continues


from track to track as long as the command is presented
to; the channel. The head number is automatically
incremented at index until the search condition is
s~tisfied, or until end of cylinder is reached. Upon

INITIAL STATUS-Normally zero.

~ 10 SEARCHED

Fie

----.J

I I I II I I I
c

(Figure 2-34)

Compares an ID in main storage with the next count ID


area on the track (including RO). If a high comparison
results, the track ID is greater than the main storage ID
and Channel End, Device End, and status modifier are
presented to the channel. If a high comparison does not
result, the track ID is not greater than the main storage
ID and Channel End and Device End only are presented
to the channel.
If a single-track search is specified, the search is
confined to one track. The search is repeated until the
search condition is satisfied or until two index points are
detected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record Found is set in sense byte 1.
If a multi-track search is indicated, the search continues
from track to track as long as the command is presented
to the channel. The head number is automatically
incremented at index until the search condition is

satisfied, or until end of cylinder is reached. Upon


detection of end of cylinder, Channel End, Device End,
and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1.
DATA ADDRESS-Specifies main storage address of a
five-byte portion of a count area containing CC, HH, and

R.
FLAGS-Used at discretion of the programmer.
COUNT-Should be five. If count is greater than five, only
the first five bytes from main storage are compared.
Channel End and Device End are presented to terminate
the command, and status modifier is presented if the
track ID was high.
If count is less than five, a comparison of main storage
and track data continues until the CCW count is zero.
Channel End and Device End are presented to the
channel when the ID and correction code bytes are read
and checked. Status modifier is presented if the search
on the short field is satisfied.
INITIAL STATUS-NormallY zero.

I - 10 SEARCHED ~
I
ON COMMAND
I

ON COMMAND

SEARCH 10 HI~H
Command Code: 51 (hex) single track
D1 (hex) multi-track

KL

DL

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

DL

EC

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

2-56

SEARCH 10 EQUAL OR HIGH


Command Code: 71 (hex) single track
F1 (hex) multi-track

(Figure 2-34)

satisfied, or until end of cylinder is reached. Upon


detection of end of cylinder, Channel End, Device End,
and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1

Compares an 10 in main storage with the next count 10


area on the track (including RO). If an equal or high
comparison results, the track 10 is equal to or greater
than the main storage and Channel End, Device End, and
status modifier are presented to the channel. If an equal
or high comparison does not result, the track 10 is less
than the main storage 10 and Channel End and Device
End only are presented to the channel.

DATA ADDRESS-Specifies main storage address of a


five-byte portion of a count area containing CC, HH, and
R.
FLAGS-Used at discretion of the programmer.
COUNT-Should be five. If count is greater than five, only
the first five bytes from main storage are compared.
Channel End and Device End are presented to terminate
the command, and status modifier is presented if the
comparison was equal or high.

If a single-track search is specified, the search is


confined to one track. The search is repeated until the
search condition is satisfied or until two index points are
detected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record Found is set on sense byte 1.

If count is less than five, a comparison of main storage


and track data continues until the CCW count is zero.
Channel End and Device End are presented to the
channel when the 10 and correction code bytes are read
and checked. Status modifier is presented if the search
on the short field is satisfied.

If a multi-track search is indicated, the search continues


from track to track as long as the command is presented
to the channel. The head number is automatically
incremented at index until the search condition is

t-I

INITIAL STATUS-Normally zero.

10 SEARCHED
ON COMMAND

--.j

I I I I I I I I I I
F

c:v
NDEX
~

G'~OME
~

G21 COUNT IG2GEY G2BATA G31

I----

RO FIELD-----...l

(11

~i7

KL

COU~T

DL

DL

EC

IG2GEY G2BATA G3

D~~A G4~NDEX
~

RECORD

~ FIELD~
RI
,AND ALL OTHERS FOLLOWINGI

AND G3 FIXED LENGTH GAPS

G4 VARIABLE LENGTH GAP

2-57

051
HEX 31.51

71
HEX BI. DI.
FI

SET DRIVE
CNTLR INTERFACE TO READ
MODE

SEARCH
ID ROUTINE
(EO, HI, EO, AND
HI)

051

READ KL, DL, DL


FROM DRIVE AND
STORE

051

(Ea. HI, EO & HI)


YES

SET NO RECORD
FOUND IN
SENSE CONTROL
BLOCK

SET END
OF CYLINDER IN SENSE

READ 10 BYTE
FROM DRIVE

051 (04F)
SET CHNL END,
DEVICE END. AND
UNIT CHECK

GET ORIENTED
TO NEXT GAP
2 OR 3

SET CHNL
INTERFACE TO
WRITE MODE

GAP 3

GAP 2

306
051

NO

051
SEND READ
GAP 3 TAG TO
DRIVE CNTLR.

SEND READ
GAP 2 TAG TO
DRIVE CNTLR.

SET SEARCH 10
PREREOUISITE IN
R3/9 TO ALLOW
CHAINED WRITES

SET CHNL END.


DEVICE END
ENDING STATUS

051 (04E)

051

051

ERRORS
POST TYPE
OF ERROR

READ CAR, HAR.


FLAG BYTES
FROM DRIVE

SET CHNL END,


DEVICE END, &
STATUS MODIFIER
ENDING STATUS

CHECK FOR
MORE ERRORS
DURING SEARCH

051
READ C, C, H, H, &
REC NO. FROM
DRIVE AND CHNL
AND COMPARE .

306

Figure 2-34. Search 10 Equal, Search 10 High, and Search 10 Equal or High Flow Diagram

3672 .21 -0001 -10/75

2-58

SEARCH KEY EQUAL


Command Code: 29 (hex) single track
A9 (hex) multi-track

(Figure 2-35)

Compares a key area in main storage with the next key


area on the track (excluding RO). If an equal comparison
results, Channel End, Device End, and status modifier
are presented to the channel. If an unequal comparison
results, Channel End and Device End only are presented
to the channel.
If a single-track search is specified, the search is
confined to one track. The search is repeated until the.
search condition is satisfied or until two index points are
de~ected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record .Found is set in sense byte 1.
If a multi-track search IS Indicated, the search continues
from track to track as long as the command is presented
to the channel. The head number is automatically
incremented at index until the search condition is
satisfied, or until end of cylinder is reached. Upon
detection of end of cylinder, Channel End, Device End,
and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1
Execution of thiS command with a key length (KL) of zero
does not set a status modifier. If followed by a chained

Read Data command, the data area read is that of the


next record.
DATA ADDRESS-Specifies main storage location where
key to be compared is located.
FLAGS-Used at discretion of the programmer
COUNT-Should be equal to KL of record containing key
to be compared. If count is greater than KL, the search
operation is completed when the key area is read.
Channel End and Device End are presented to the
channel, terminating the command. Status modifier is
presented if the comparison was valid.
If count is less than KL, the track and main storage
comparison continues until the CCW count is zero.
Channel End and Device End are presented after the key
area and subsequent correction code bytes are read and
checked. Status modifier is presented if the search on
the short field was satisfied.
INITIAL STATUS-Normally zero.
SPECIAL NOTE-When command is chained from
Search ID or Read Count, the key compared IS in the
same record as the 10 or count. Search Key Equal
bypasses RO key field unless chained from Search ID
command, which search RO.

SEARCH KEY HIGH


Command Code: 49 (hex) single track

(Figure 2-35)

Compares a key area In main storage with the next key


area on the track (excluding RO). If a high comparison
results, the track key is greater than the main storage key
and Channel End, Device End, and status modifier are
presented to the channel. If an equal or high comparison
does not result, the track key is not greater than the main
storage key and Channel End and Device End only are
presented to the channel.

If a single-track search is specified, the search is


confined to one track. The search is repeated until the
search condition is satisfied or until two index points are
detected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record Found is set in sense byte 1.

If a multi-track search is indicated, the search continues


from track to track as long as the command is presented
to the channel. The head number is automatically
incremented at index until the search condition is
satisfied, or until end of cylinder is reached. Upon
detection of end of cylinder, Channel End, Device End,
and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1.

Execution of thiS command with a key length (KL) of zero


does not set a status modifier. If followed by a chained
Read Data command, the data area read is tha't of the
next record.
DATA ADDRESS-Specifies main storage location where
key to be compared IS located.
FLAGS-Used at discretion of the programmer.
COUNT-Should be equal to KL of record containing key
to be compared. If count is greater than KL, the search
operation is completed when the key area is read.
Channel End and Device End are presented to the
Channel, terminating the command. Status modifier is
presented if the track key was high.
If count is less than KL, the track and main storage
comparison continues until the CCW count is zero.
Channel End and Device End are presented after the key
'area and subsequent correction code bytes are read and
checked. Status modifier is presented if the search on
the short field was satisfied.
INITIAL STATUS-Normally zero.
SPECIAL NOTE-When command is chained from
Search ID or Read Count the key compared is in the
same record as the 10 or count. Search Key Equal
bypasses RO key field unless chained from Search ID
command, which searched RO.

SEARCHED ON COMMAND

r=:-tJ~::lG21
'V
LJ

I
0

COUNT IG21:lG2EJATA G31 COUNT IG21:lG2EJATA G3

~~~ G4~
RECORD

I
ROFIELD----..l
~RIFIELD~
I---(AND ALL OTHERS FOLLOWING)

Gl. G2. AND G3: FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

Gl. G2. AND G3 FIXED LENGTH GAPS


G4: VARIABLE LENGTH GAP

2-59

SEARCH KEY EQUAL OR HIGH

(Figure 2-35)

Command Code: 29 (hex) single track


A9 (hex) multi-track
Compares a key area in main storage with the next key
area on the track (excluding RO). If a high comparison
results, the track key is greater than the main storage key
and Channel End, Device End, and status modifier are
presented to the channel. Ifan equal or high comparison
does not result, the track key is not greater than the main
storage key and Channel End and Device End only are
presentee to the channel.

If a single-track search is specified, the search is


confined to one track. The search is repeated until the
search condition is satisfied or until two index points are
detected. Upon detection of the second index, Channel
End, Device End, and Unit Check are presented to the
channel; and No Record Found is set in sense byte 1.

If a multi-track search is indicated, the search continues


from track to track as long as the command is presented
to the channel. The head number is automatically
incremented at index until the search condition is
satisfied, or until end of cylinder is reached. Upon
detection of end of cylinder, Channel End, Device End,
and Unit Check are presented to the channel; and End of
Cylinder is set in sense byte 1.

Execution of this command with a key length (Kl) of zero


does not set a status modifier. If followed by a chained
Read Data command, the data area read is that of the
next record.
DATA ADDRESS-Specifies main storage location where
key to be compared is located.
FLAGS-Used at discretion of the programmer.
COUNT-Should be equal to Kl of record containing key
to be compared. If count is greater than Kl, the search
operation is completed when the key area is read.
Channel End and Dev!ce End are presented to the
channel, terminating the command. Status modifier is
presented' if the comparison was equal .or high.
If count is less than Kl, the track and main storage
comparison continues until the CCW count is zero.
Channel End and Device End are presented after the key
area and subsequent correction code bytes are read and
checked. Status modifier is presented if the search on
the short field was satisfied.
INITIAL STATUS-Normally zero.
SPECIAL NOTE-When command is chained from
Search /D or Read Count the key compared is in the
same record as the ID or count. Search Key Equal
bypasses RO key field unless chained from Search ID
command, which searched RO.

Gl. G2. AND G3; FIXED LENGTH GAPS


G4; VARIABLE LENGTH GAP

2-60

051

052 (06DI

HEX 29, 49,

69
HEX A9, C9,
E9

CLOCK
NEXT DATA
FIELD

SEARCH
KEY ROUTINE
(EQ, HI, EQ & HI)

SET SEARCH
KEY PREREQUISITE IN R3/9

POST TYPE
OF ERROR

052 (04E)

052
SET DRIVE CNTRL
TO READ MODE
AND CHNL IIF TO
WRT MODE

052

YES

SET CHNL END


AND DEVICE
END

SET CHNL END


AND DEVICE
END

CHECK FOR
MORE ERRORS
DURING SEARCH

052

READ 10 BYTE
FROM DRIVE

GET ORIENTATION
TO NEXT COUNT
FIELD (EXCLUDING
ROi

SET CHNL END.


DEVICE END. AND
STATUS MODIFIER

YES

051 (04F)

052 (041)
READ 10, CAR
HAR, FLAG, C, C,
H, H, R, KL, DL,
DL BYTES FROM
DRIVE AND STORE
IN BUFFER

052
SEND READ
GAP 2 TAG
TO DRIVE
CONTROLLER

306

READ NO OF KEY
BYTES AS
SPECIFIED BY
KEY LENGTH IN
R1/C

052
COMPARE KEY
BYTES FROM
DRIVE AND FROM
CHNL

SET END OF
CYLINDER
IN SENSE

SET NO RECORD
FOUND IN
SENSE CONTROL
BLOCK

051 (04F!

SET CHNL END,


DEVICE END.
UNIT CHECK

Figure 2-35. Search Key Equal. Search Key High. and Search Key Equal or High Flow Diagram

3672 ,21 -0001-10175

2-61

2.9 SELECTIVE RESET


The Selective Reset function is described by means of
the flowchart in Figure 2-36. Refer to format 3 portion of
Table 2-5 for bit definitions of the Selection Reset sense
bytes.
Machine Check Multiplex (MCM) is set to obtain data
from the machine. Check 1 conditions are multiplexed
through the Machine Check Register (MCK). When
assembling of sense bytes is completed, Check 1 and 2
conditions are reset.
MACHINE HARDWARE
STARTS tLPROG
AT LOCATION 000
AND SET MCK
REG BIT 3

07C
PERFORM
CONTROLLER AND
DRIVE RESET
AS IF SELECTED

ASSEMBLE
SENSE BYTES
8 THROUGH 23
(PARAGRAPH 2.5.3)

INITIALIZE SENSE
CONTROL BLOCK TO
SHOW UNIT CHECK
IS OUTSTANDING
FOR SELECTIVE
RESET

07C
BRANCH TO
BASIC
WAIT LOOP

302

Figure 2-36. Selective Reset Flow Diagram

':l~7?

? Lnnn1 _10/7'"

2-62

2.10 SYSTEM RESET


Figure 2-37 illustrates the sequence of operations during
a System Reset. In addition to the conditions described in
paragraph 3.5.17, System Reset also occurs as follows
301

When the power for the system is turned on.

SET OLD DEOS


AND PCH DEO'S
INTO
NEW DDW 0

YES

When the RESET IL T switch is pressed.

When the channel is off-line to the interface.

As part of the initial program loading procedure.

301 (03BI
RESET
CHECK 1
AND
CHECK 2
ERRORS

RESET INDEX
AND
SELECT
DRIVE

301
OR' BSDA
OF THIS DRIVE
INTO ACCUM
DER REG

301 (03B)
RESET INTS AND
CONTROL RST
BOTH CNTRLR
AND DRIVE

301

SET POLL
ENABLES IN
CHNL INT WORD
FOR OTHER
3 CHNLS

301

301

RST DER'S & DCC'S


IN CHNL INTRPT
WORD FOR ALL
STRINGS & CHNLS

SET DER'S
IN CHNL INT
WORD FOR
OTHER 3 CHNLS

RESET DEO'S
FOR
RESETIING
CHNLS

301

301

301

CLEAR
SCRATCH PAD
PAGES
O-F

STORE DDW 0
AND DDW 1
BACK IN
BUFFER

DECREMENT
STRING
POINTER

301

301

GET
DRIVE
STATUS

DROP
DRIVE
SELECTION

DECREMENT
LOGICAL DRV
ADRS TO SET
ADRS = 7

DECREMENT
BSDA

Figure 2-37. System Reset Flow Diagram

3672.21 -0001-10175

2-63

2.11 COMMAND RETRY


Command Retry is a channel/SCU procedure that causes
automatic retry of an improperly executed command in a
channel program. The retry does not cause an 1/0
interrupt, and programmed error recovery procedures are
not required.

2.11.1 Check and Error-Initiated Retry

If a correctable error occurs In a subsequent overflow


segment In the Data field, Key field, Count field, or Home
Address field, the error is corrected by the system ERP
using information sent to the channel in the sense data.

ment Check sense information to be presented to the


channel for use In the Retry Command Check

2.11.5 Write Offset

2.11.2 Defective or Alternate Track

Write Offset occurs when the Write command detects


that the access is offset. Write Offset causes the
command to be retried after offset is reset

2.11.1.2 SYNC CHECK

When an alternate or defective track condition IS


detected before data transfer begins, the SCU determines
the location of the alternate or defective track (from RO
on the track), initiates a seek to this track, orients on
index, and reissues the original command.

2.11.6 Invalid Count Field Sync Detected

A Sync Check is executed when a sync byte is missed


during a read operation. The Uncorrectable Flag is set in
the microprogram to signify that the error is uncorrectable. The sequence then is the same as for a Data
Check.

2.11.1.1 DATA CHECKS


If a correctable error occurs in the Key field, Count field,
or Home Address field and the error is not in a
subsequent overflow segment, the command is retried.
The Key, Count, or Home Address read from the disc is
placed in a buffer in the SCU. When the correctable error
occurs, the SCU corrects the data in the buffer and
requests the channel to reissue the command which
originally caused the error. During reorientation to the
record, the SCU disconnects and frees the channel.
When the failing command is reexecuted, the corrected
data in the buffer is used instead of the actual data from
the track
If a correctable error occurs in the Data field and the
error is not in a subsequent overflow segment, the error
is corrected by the system Error Recovery Procedure
(ERP) using information sent to the channel in the sense
data.
If, during a Read operation, an uncorrectable error occurs
in the Key field, Count field, or Home Address field and
the error is not in a subsequent overflow segment, the
Read command is retried a maximum of 28 times. If the
retry is not successful, the error becomes a permanent
error. If this error occurs during a Write operation, the
Write command is retried a maximum of 15 times.

2.11.1.3 AM CHECK
An AM Check is executed when the Address Mark is
missed on retry. The Address Mark error is uncorrectable. The sequence of the AM Check is the same as that
given for an uncorrectable error of a Read operation that
does not occur in a subsequent overflow segment.

2.11.1.4 PHYSICAL 10 ERROR - SEEK CHECK


Physical 10 Errors are detee-ted during read-end processing of the Count Buffer against the Seek CAR-HAR in the
Drive Data Buffer. If the Physical 10 Error is not in an
overflow segment, the Seek is retried a maximum of ten
hmes. If the retry is not successful, the error becomes a
permanent error. If the Physical 10 Error occurs in an
overflow segment, Operation Incomplete sense information is presented to the channel in addition to status of
Channel End, Device End, and Unit Check.

2.11.1.5 SEEK INCOMPLETE ERROR


Aher initial status has been presented, Seek Incomplete
errors are detected for all untimed commands, except
Sense commands. Seek Incomplete errors are retried
and, if interrupted or incomplete on retry, cause Equip-

2.11.3 Command and Data Overrun


When a command overrun or late command chaining
condition occurs because of interference from another
channel or the CPU, the SCU initiates a retry of the
command that was late.
When a data overrun occurs, a retry of the command is
initiated, except when the data overrun occurs during
one of the following operations:

During a record overflow operation in the subsequent overflow segments.

During a format write operation.

Invalid Count field sync is detected when the index


occurs prior to the third ECC byte of the Count field and
after the Lock VFO to Data Point in the gap preceding the
Count field. Invalid Count field sync causes the command
to be retried.

2.11.7 Index Detected in Data Field


Index is detected in the Data field when the index occurs
between the Lock to VFO Point and the Source 10 Byte of
the Data field. Index detected in the Data field causes the
command to be retried.

2.11.S Command Retry Flowcharts


2.11.4 Padding
Padding occurs when the Write gate is on after initial
status presentation and the command is not a Format
Write. The padding sequence tests to see if the drive has
the disconnected auto padding feature present. If the
feature is active, the SCU presents retry status and lets
the drive disconnect from the SCU to finish padding and
reconnect as if it were a disconnected command chaining sequence. If this feature is not installed in the drive,
padding will cause the command to be retried and the
SCU to wait for index before reconnecting to the
command.

Figures 2-38 through 2-45 which follow contain


flowcharts for the Command Retry procedures discussed
in the preceding paragraphs. The flowcharts are
presented in the following order:

Data, Sync, and AM Checks (Figure 2-38)


Physical 10 and Seek Errors (Figure 2-39)
Defective Track (Figure 2-40)
Alternate Track (Figure 2-41)
Data and Command Overruns (Figure 2-42)
Padding (Figure 2-43)
Write Offset (Figure 2-44)
Invalid Count Field Sync Detected and Index
Detected in Data Field (Figure 2-45)

2-64

071
DECODE ECC
POLYNOMIAL TO
DETERMINE
CORR PTIRN
AND DSPLCMT

SET CHNL
FREEZE AND
RESET CHNL
TRANSFER

SET
FORMAT/MESSAGE
IN SCB

SET SCB =
OP INCOMPLETE
AND
CORRECTABLE

RESET
CORRECTABLE
INDICATOR
IN SCB

SET SCB =
CORRECTABLE

071

071

071

071

071 (030)
RESET
BUFFER VALID
FOR HA, CNT,
OR KEY FIELDS

UPDATE ORIENT
REG TO REFLECT
FIELD THAT WAS
IN ERROR

RESET
R/W
CONTROL

071
071 (07G)
SET
UNCORRECTABLE
INDICA TOR AND
RESET
TRUNCATION

SET
FORMAT
= 150

UPI;>ATE
RETRY
COUNTERS
SET
CORRECTABLE
IN SCB

071

SET
COUNT
ORIENTATIO(l;

PRESENT
RETRY
STATUS

SET
FORMAT
40

071 (07E)

071

SET
DATA FIELD
IN SCB

PERFORM
OFFSET
PROCESSING

CORRECT
BUFFER

071

SET
"DOUBLE
ORIENT"

071 (07A)

SET
OP INCOMPLETE
IN SCB

SET
KEY FIELD
IN SCB

071 (03H)

GET
RETRY
SECTOR

071 (07A)

PRESENT
RETRY
STATUS

071 (07H)

LOG
ERROR

071

071 (07H)
PERFORM
LOGGING
IF
REQUIRED

PRESENT
UC ON INITIAL
STATUS OF
RETRIED CMD

071 (076)
PERFORM
ORIEIlIJATION
TO RETRY
CMD

071
SET
COUNT FIELD
IN SCB

If 2860 Attachment Feature


stalled. See paragraph 3.12

In

Figure 2-38. Command Retry, Data, Sync, and AM Checks


3672.21-0001-10175

2-65

Detected dunng read-end processing of

Detected for all commands except sense


after untimed initial status has been presented.

count buffer against the seek CAR-HAR


in the drive data buffer.

073 (07G,
YES

INCREMENT
SEEK RETRY
COUNT
073

073 (07AI

SCB=/5700
OP INCOMPL

PRESENT
RETRY STATUS

073 (07AI
SENSE CTRL
BLK=/031A.
EQUIPMENT
CHECK. SEEK
ERROR. AND
PERMANENT
ERROR

r--- ---,
I
SCU
It

~
I
_ __ J

DISCONNECTS

I FROM CHANNEL

L___

073 (03E)

073 (07B)

STATUS"CE.
DE.AND UC

RECALIBRATE
START

073 (03GI
BIT SIGNIFICANT POLL FOR
DRV INTRPT
306

If 2860 Attachment Feature Installed See paragraph 3.12

RST
INTRPT

073 (03A)

SEEK TO
PROPER TRACK

073 (03G)
BIT SIGNIFICANT POLL FOR
DRVINTRPT

YES

Figure 2-39. Command Retry Physical 10 and Seek Errors

3672 .21 -0001-1 0175

(1

TO
(2 OF 2)

TO
(2 OF 2)

of 2)

2-66

FROM
(1 OF 2)

..

073 (078)

PRESENT UNTIMED
INITSTATUS=O
. REQUEST
RECONNECTION

073 (078)

r-- --,
I
I

scu
RECONNECTS
TO CHANNEL

L.__

It
I
I

073 (078)

PRESENT
DEVICE END

073 (078)

073 (03G)

SET
ZEROTRK
ORIENTATION

073 (03G)

SCB=/OB1C
EQUIPMENT
CHECK FMT 1
NO INTRPT
FROM DRV

I
~

073 (03G

078

REQUEST
RECONNECTION

073 (071)
ORIENTTO
INDX AND
SET INDX
ORIENTATION

(1FROM
OF 2)

SAVE
CUDI STATE

CONTROL
RESET
TO DRIVE

__..J

078

SAVE
CUDI STATE

073 (03G)

CONTROL
RESET

SELECTION

TO DRIVE

078

PRESENT
DEVICE END

078

073

SCB=/OB1B
EQUIPMENT
CHECK FMTl
SEEK INCOMPL
ON RETRY

CHAINED
RESELECTION

CHAINED
RESELECTION

073 (078)

073 (078)

RETRY
CMD CHECK

RETRY
CMDCHECK

073 (078)

INIT STATUS =
UNIT CHECK

I
END STATUS

If 2860 Attachment Feature


stalled. See paragraph 3 12

306

Figure 2-39. Command Retry, Physical 10 and Seek Errors (2 of 2)

2-67

3672 .21 -0001-10175

074 (07AI

074
SCB=/OOOB
CMD REJECT
FMTO
IMPROPER
ALTTRK
POINTER

PRESENT
RETRY STATUS

074 (306,

078

r-- - - ,
I

scu

DISCONNECTS
FROM CHANNEL

SCB=/5700
OP INCOMPL

FROM
(2 OF 2)

I t

SCU
RECONNECTS
TO CHANNEL

L _ _ _ _ .J
074 (041)

074

STATUS=CE,
DE,AND UC

r--

--1

I
__J

078

I SCU RECONNECTS I

GET RO
COUNT
ORIENTATION

TO CHANNEL

L__

078

YES

PRESENT
DEVICE END

078

SEEK TO
ALTERNATIVE

CHAINED
RESELECTION

306

074 (03G)

BIT SIGNIFI
CANT POLL FOR
DRV INTRPT

If 2860 Attachment Feature in


stalled. See paragraph 3.12.

PRESENT
UNIT CHECK
FOR
INITIAL STATUS

TO
(2 OF 2)

Figure 2-40. Command Retry, Defective Track (1 of 2)

3672.21-0001-10/75

2-68

074 (03G)
TIME OUT
FORINTRPT
OF 780 MSEC
RST INTRPT

074 (03G)

SAVE CUDI
STATE

074 (07B)
REOUEST
RECONNECT

r-I
I
I

SAVE
CUDI STATE

074 (03G)
CONTROL
RESET
TO DRIVE

-- ....

SCU RECONNECTS
TO CHANNEL
(ON INTERRUpn

L ____

074 (03G)

It
I

074 (03G)
CONTROL
RESET
TO DRIVE

074

SCB=/OB1B
EOUIPMENT
CHECK FMT1
SEEK INCOMPL
ON RETRY

074 (07B)

PRESENT
DEVICE END

074
SCB=/OB1C
EQUIPMENT
CHECK FMT 1
NOINTRPT
FROM DRV

074 (07B)

CHAINED
RESELECTION

074 (07B)

074 (05F)

PRESENT ZERO
INITSTATUS
TO

RETRY
CMD CHECK

074

(1 OF 2)

GETINDX
ORIENTATION

If 2860 Attachment Feature installed. See paragraph 3.12.

Figure 2-40. Command Retrv. Defective Track (2 of 2)

3672.21-0001-10175

2-69

074 (07A)

PRESENT RETRY
STATUS

NO

074 (041)

r-- --.,
I

scu

I DISCONNECTS I
I FROM CHANNEL I
L _ _ _ _ .J

GET RO
COUNT
ORIENTATION

074
074

GETRO
COUNT

INCREMENT
HEAD
ARGUMENT

NO
NO

074

SCB=/08OB
CMDREJECT
FMTO
IMPROPER
ALTTRK
POINTER

074

074

074
SCB=/5200
OP INCOMPL
END OF CYL

SET
HEAD ZERO
SCB=/5700
OP INCOMPL

SET HEAD=
HAR+l

078
NO

1>74 (03A)
SEEK (TO
DEFECTIVE

SCB=/1200
END OF CYL

+1)

306
If 2B60 Attachment Feature installed. See paragraph 3.12.

TO

(2 Of 2)

TO
(2 OF 2)

. Figure 2-41. Command Retry, Alternate Track (1 of 2)

3672 .21 -0001 -10175

2-70

SAVE
CUDI STATE

SAVE
CUDI STATE

074 (03G)

CONTROL
RESET
TO DRIVE

CONTROL
RESET
TO DRIVE

RST INTRPT

074 (07B)

074

074

SCB={OB1B
EQUIPMENT
CHECK FMT 1
SEEK INCOMPL
ON RETRY

SCB={OB1C
EQUIPMENT
CHECK FMT 1
NO INTRPT
FROM DRV

FROM
(1 OF 2)

074 (03G)

074 (03G)

REQUEST
RECONNECTION

If 2860 Attachment. Feature installed. See paragraph 3.12.

r--

078

I .
I

REQUEST
RECONNECTION

.-I
I
I

--...,

SCU RECONNECTS
TO CHANNEL
(ON INTERRUpn

L...__

I t
I
I

__.J
074 (078)

--,

SCU RECONNECTS
TO CHANNEL
(ON INTERRUPT)

I t

RETRY
CMDCHECK

L- _ _ _ _ ..J

PRESENT
DEVICE END

074 (078)

PRESENT
INITSTATUS
OF ZERO

074

078

078

PRESENT
DEVICE END

074 (05F)

078

PRESENT
UNIT CHECK
INITSTATUS

CHAINED
RESELECTION

GET INDX
ORIENTATION

074 (078)

078
RETRY
CMDCHECK
CHAINED
RESELECTION

306

Figure 2-4l. Command Retry, Alternate Track (2 of 2)

2-71
3672.21-0001-10/75

075 (07B)

PRESENT
DEVICE
END

075 (07B)

075 (07G)

INCREMENT
DATA OVERRUN
RETRY
COUNTER

SET SCB
PERMANENT
DATA OVERRUN
ERROR

078

YES

CHAINED
RES ELECTION .

075 (07A)

RETRY
CMD
CHECK

PRESENT
RETRY
STATUS

STATUS =
CEo DE. AND UC

075

SET SCB
OVERRUN

075 (05F)

I
306

SCU - - - ,
DISCONNECTS
FROM
CHANNEL

II

t
PRESENT
ZERO INIT
STATUS

----1

L_

PAD
TO
INDEX

075 (071)
ORIENT
TO
COARSE
SECTOR

r-I

INCREMENT
DATA
OVERRUN
COUNTER

075 (07B)

FINE
TRACK
ORIENTATION

--,
SCU
RECONNECTS
AT COARSE
SECTOR

L_

I
I

_--'

If 2860 Attachment Feature installed. See paragraph 3.12.

Figure 2-42. Command Retrv. Data and Command Overruns (1 of 2)

3672 .21 -0001-10175

2-72

..

_-

--

_._---,

YES
SET
CMD OVERRUN
ON STD IN
R2IB

NO

SET DOUBLE
ORIENT TO
OBTAIN NEXT FLD

SET
COUNT
ORIENT

075
RST
CURRENT
SECTOR
FROZEN

PAD
TO
INDEX

075 (07G)

PAD UNTIL
INDEX
AND READ
Gl AGAIN

INCREMENT
CMD
OVERRUN
COUNT

I
I

'-------.~

Figure 2-42. Command Retry, Data and Command Overruns (2 of 2)

3672.21-0001-10175

2-73

Write command detected that access was


offset.

076
RESET
CURRENT
SECTOR
FROZEN

048 (03D)
WAIT
100 pSEC
FOR INDEX
TO COME UP

PRESENT
RETRY
STATUS

ARM
PADDING

077
RESET
INDEX PASSED
INDICATOR

048 (07A)

076 (07B)

PRESENT
RETRY STATUS

077 (04A)

WAIT
48 MSEC
FOR INDEX

PRESENT
RETRY
STATUS

077 (07B)

04B

r-scu

DISCONNECTS
FROM CHANNEL

CHAINED
RESELECTION

PRESENT
DEVICE END

076 (07B)

--1

I
L.-. _

RESET
FMT WRITE
IN
PROGRESS

r-- ---,
I SCU RECONNECTS I
I TO CHANNEL I
I (ON INTERRUPT) I
L ____ ...I

I
I
I

CHAINED
RESELECTION

_ _ ....

076 (03J)
076 (07B)
READ TARGET
CTRL RSTOFFSET

RETRY
CMDCHECK

077 (07B)

04B

076 (07D)
READ
DEVICE
STATUS

PRESENT
ZERO
INITIAL
STATUS

SET UP INDICS
TO TREAT
DE AS A
DCC

076 (05F)

COARSE SCTR
ORIENTATION

PERSENT
ZEROINIT
STATUS
076 (031)
REQUEST
RECONNECT
AT COARSE
SCTR

076 (071)

FINE TRK
ORIENTATION

306

If 2860 Attachment Feature installed.


See paragraph 3.1 2.

Figure 2-43.. Command Retry, Padding

3672.21 ~OOOl-l 0175

If 2860 Attachment Feature installed. See paragraph 3.12.

Figure 2-44. Command Retry, Write Offset

2-74

Index occurred prior to third ECC byte of Count field and after the Lock
VFO to Data Point in the gap preceding the Count field. This is caused by
writing CKD near index or when the write turnoff glitch under index
looks like a valid count field sync byte and happens to occur the correct
distance from the previous data field. The error occurs on the subsequent
read. When this error is detected, ECC is not processed.

072 (07A)

PR ESENT
RETR Y STATUS

r-I
I

I t
I

..... -- T--~
TA RGET

072 (07B)

RETRY
CMDCHECK

seTR SET TO

zERO

072 (07D)

072 (04A)

ORIENTTO
INDX. DO
NOT RST INDX

CO ARSE
SCTR
ORIE NTATION

072 (07B)

RE-ENTER CMD)

REC ONNECT
AT COARSE
SCTR

1.. __ ,

II SCU RE CONNECTS:

I
L.. _ _ ____ .J
:

072 (07B)

CHAINED
RESELECTION

072

r--

072 (07B)

PRESENT
DEVICE END

1__ -,

SCU
DISC ONNECTS
FROM CHANNEL

. . - Index occurred between the Lock to VFO Point and the Source ID Byte of
the Data field because the Data field is being clocked at a virtual position
offset to provide more channel turnaround in Gap 2 and the offset virtual
data field overlaps index.

TOC HANNEL
(ON IN TERRUPT)

t
t

If 2860 Attachment Feature


stal led. See paragraph 3.12

In

Figure 2-45. Command Retry, Invalid Count Field Sync Detected and Index Detected in Data Field

3672.21-0001-10175

2-75

This page intentionally left blank.

3672.21-0001-10175

2-76

2.12 ROTATIONAL POSITION SENSING


2.12.1 Introduction
Rotational position sensing (RPS) is a subsystem feature
that allows the channel and the SCU to be released
during most of a record search time, thereby increasing
their availability for other operations. Two commands,
Read Sector and Set Sector. are associated with RPS
operation. These commands are described in the following paragraphs

READ/SET SECTOR OPERATION


Read/Set Sector operation is performed as follows:

SET SECTOR GIVEN

OEVICE END

CHANNEL RESELECTION DELAY

RECORD READ
OR WRITE

I nit i al Selection
Channel End

Sector number, relative


to the readfwrite
heads, adjusted for

If channel does not


respond, connection is
tried on subsequent

If channel reconnection is established, the


record is ready to be

Disconnect

reselection delay.

revolutions.

read or written.

2.12.2 Read Sector


The Read Sector command transfers one byte of information from the SCU to main storage (Figure 2-46) This
byte provides the sector number (angular displacement
from index) of the last record processed. If the last record
processed was an overflow record. the sector number
returned is that of the first segment processed in the
current command chain.
Execution of a Read Sector command causes loss of
record orientation previously established. Therefore, if
further processing of records of the same track (or
cylinder) is desired, a search for the desired record must
be made.

2.12.3 Set Sector


Refer to Figure 2-46. The Set Sector command transfers
one byte of data from main storage to the SCU. ThiS byte
specifies one of 128 possible sector numbers per track.
The byte value is checked for validity by the SCU. If the
value is proper, the SCU generates Channel End and
allows the channel to disconnect.
When the desired sector (which has an adjusted reselection delay factor) is reached, the SCU signals Device End.
The channel can then connect to contin'ue the command
chain. If the channel does not respond within 130
microseconds, connection is tried on subsequent
revolutions.

Figure 2-46 Read/Set Sector Operation

2-77

2.12.4 Disc and Track Layout


The track layout on individual recording surfaces in the
disc pack is shown in Figure 2-47. A recording surface is
divided into 128 sectors (0-127). The time for pack
rotation through one sector IS approximately 130
microseconds.
When a record is written, the sector number may be
stored in the using system by a Read Sector command.
For example, assume Record 3 is written. Referring to
Figure 2-48, Sector 9 is stored by using the Read Sector
command. This sector number can be calculated by the
using system if the record lengths are fixed.
The sector number is set in the Target Register of the
selected drive. The sector that is set (target sector) is
always two less than the desired sector. When the sector
counter and Target Register compare, SCU raises ReQuest In to the channel. Then the channel can command
a Search 10 for the desired record with minimum loss of
system/CPU time.

Figure 2-47. Track Sector Layout

123

124

125

126

127

Figure 2-48. Typical Track Field Layout

2-7R

2,13 MULTITRACK OPERATION


The flowchart In Figure 2-49 illustrates the sequence for
a multitrack Read or Search operation A multitrack
coded Search or Read command must be executed to
perform
multitrack
operation.
During
multitrack
operations, the drive Head Address Register (HAR) is
automatically incremented, thus eliminating the need for
Seek Head commands in a chained Read or Search
command.

Bit 0 of command code byte

IS

a "1".

YES

Select a single track Read HA or Read


RO.

(SEE PARAGRAPHS
2.6 AND 2.8)
YES

WAIT FOR
INDX

INCR HAR
TO DRVAND
KEEP INDX
ORIENTATION

YES

SET END
OF CYL
SENSE

Figure 2-49. Multitrack Operation

2-79

2.14 OVERFLOW RECORD

Search 10 (RO)

2.14.2 Formatting

Write Special CKD (3rd segment)


Seek Head (next track)

2.14.1 Introduction
These segments are written using Write Special Count.
Key, and Data commands. The sequence IS as follows
(letters refer to corresponding tracks of the record):
Overflow Records allow logical data records longer than
one track to be read or written on the disc. Format
writing of an overflow record is controlled by the using
system. Read and nonformatting write operations are
controlled by the SCU.

Search 10 (RO)
Write CKD (4th segment)

2.14.3 Processing
Search 10 (R1)
Write Special CKD (1 st segment)
Seek Head (next track)

The SCU switches to the next head on all overflow


record~ (flag byte bit 4 = 1) for read and nonformatting
write commands. The command sequence consists of a
Search 10 (R2) succeeded by one of the following: Read
Key and Data. Read Data. Write Key and Data. or Write
Data (read or write segments 1. 2. 3. and 4 with auto
head switching at index. controlled by the SCU).

Search 10 (RO)

Formatting and processing of Overflow Records are


described in Figure 2-50 and the following paragraphs.

Write Special CKD (2nd segment)


Seek Head (next track)

Errors on the first segment (excluding correctable data


field data checks) are retried by the SCU.

An error on the second or subsequent segments (data


check. defective track. etc.) sets sense byte 1. bit 7
(Operation Incomplete). A restart CCW is provided to the
using system error recovery procedure (ERP) in sense
byte 3. After the ERP has corrected the problem. the
restart CCW is sent to the SCU to resume the operation.

00000110 (06) A read command was in progress.


00000101

(05) A write command was in progress.

.tl

--liHo....- - - - - - - - - - - - - FIRST SEGMENT OF OVERFLOW R E C O R O - - - - - - - - - -_ _ _ _ _ _..

INOEX
R2-0 (SEGMENT 11

ffi

r-

~~

FLAG BYTE BIT 4 = 1

+[HAl R R B ~IONAL~------------------------------------------------------------------------------------------_,
EJ I
+
R1-0 (SEGMENT 21

(2l or FLAG BYTE

, [HAl B

BIT 4 = 1

----------------------------~----------------------------_,
R~,~B OPTIONAL~-----------------------------8 I
__

R1-0 (SEGMENT 31

(0). ~ FLAG BYTE


'-'"

BIT 4 = 0

OPTIONAL~----------------------------------------.......------...,
R1-0 (SEGMENT 41

OPTIONAL,,-____________________________- ,
R2':"O

14------------- LAST SEGMENT OF OVERFLOW R E C O R D - - - - - - - - . ,

Figure 2-50. Overflow Record Operation

2-80

TABLE 2-6. DEVICE RESERVE/RELEASE COMMAND FUNCTIONS

2.15 MULTICHANNEL OPERATION


Multichannel operation permits the SCU to communicate
with up to four CPUs over any of four channels
(maximum/ Such operation allows any of the eight disc
drives controlled by the SCU through the Controller to be
reserved exclusively for any of the channels so enabled.
A channel is enabled for multichannel operation by
means of a corresponding CHANNEL switch on the SCU
Operator Control Console. There are four switches. one
for each channel. Setting the switch to the enable
position merely enables the channel for operation. actual
channel switching and device reservation IS controlled by
the channel program.

COMMAND

HEX
CODE

Device
Reserve

B4

Device
Release

94

FUNCTION
1

2.
3.

1.
2.
3.

ERROR TYPE

Set Bit 0 of RO. page 9 of Register FIt.: ,n micro


processor.
Transfer 24 bytes of sense data to channel.
Send ending status to channel.

If command IS preceded by a Set File Mask In the chain.


Command Reject will be set In sense data and Unit Check will
be initial status.

Reset Bit 0 of RO. page 9 of Register File in microprocessor.


Transfer 24 bytes of sense data to the channel.
Send ending status to channel.

Reservation of a device is accomplished by executing a


Device Reserve command. A device address reserved to
one channel. or being used by one channel. presents a
Busy status to any other channel if selection is
attempted. Release of a device previously reserved is
accomplished by executing a Device Release command.
In addition. 24 bytes of sense information are transferred
to the channel when either the Device Reserve or Device
Release command is executed. (Refer to paragraph 3.11
for a detailed description of multichannel switch operation.)

--

--

STORAGE CONTR

MICROPROCESSOR

CONTROLLER INTERFACE

L=d====d=
I
L

---r- -

CONTROLLER

--

.-------------~

CPU

DISC DRIVE
MODULE 3

DISC DRIVE
MODULE 4

PHYSICAL
ADDRESS
110001

PHYSICALa

~l~~~lESS~

101010

DRIVE D

PHYSICAL
ADDRESS
100011

~
~

t.J ______ ~

I
I

1_

DRIVE G

DRIVE E

PHYSICAL~
ADDRESS

i
~

DISC DRIVE
MODULE 1

DRIVE C

PHYSICALti}
ADDRESS
111000

DRIVE B

DISC DRIVE
MODULE 2

I
DRIVE A

CHANNEL A

CHANNEL INTERFACE

I
I
I

The block diagram In Figure 2-51 Illustrates multichannel


operation considering two channels. An associated table
for the Device Reserve and Device Release commands
are presented in Table 2-6. Refer to Figure 2-16 for the
corresponding flowchart of these commands.

-1-

-- -

i
DRIVE F

PHYSICAL~
ADDRESS
011100

___

I
I

.1..~

PHYSICAL~

ADDRESS
001110

r--'-----,
DRIVE H

PHYSICAL~
I
ADDRESS
000111

______ L--1

Figure 2-51. Multichannel Operation Block Diagram

2-81

SECTION 3. CHANNEL INTERFACE DESCRIPTION

II

3.1 GENERAL
This section describes the I/O interface between the
System/370 channel and the 3672 Storage Control Unit
(SCU). In general, this interface is characterized by the
following features:

Provides signal sequence and information flow


common to all SCUs.

Is defined by IBM, and amended by Memorex for


the 3672 SCU in terms of interface line
groupings and additional special controls.

P.O. 1,2,3,4,5,6,7

BUS OUT LINES


NINE BIT POSITIONS
p. 0, 1,2, 3,4, 5,6, 7

Provides for common connection and communication between an IBM System/370 block
multiplex channel and SCU.

II

1\'f
SYSTEM

370

Logically connects only one SCU to the channel


at a time.

The standard channel Interface is composed of 37 lines


which are divided into five groups: bus lines, tag lines,
scan controls, interlock lines, and special controls. These

V
~

I\J

I\r

vi

Tag lines are used for special sequences and for


interlocking and controlling information on the buses.

,J
-\

-V

Scan controls enable an SCU to request service from the


.channel on a priority basis, These controls also allow the
channel to contact the SCU,
3672
STORAGE
CONTROL
UNIT

Interlock lines lock out nonselected SCUs, permitting


only one SCU to be serviced at a time, The interlOCk lines
also gate the tag lines, and when down, initiate reset of
the interface operations.
Special controls are used to provide special functions:
conditioning of usage meters, CPU interlock control, and
a marker tag line for command retry.

INTERLOCK LINES IN:


OPERATIONAL IN

INTERLOCK LINES OUT:


OPERATIONAL OUT

-\

SCAN CONTROLS IN
SELECT IN
REOUEST IN

SCAN CONTROLS OUT


SELECT OUT
HOLD OUT

l;t

Bus lines transfer information between the SCU and the


channel. Each bus is a set of nine lines consisting of one
parity line and eight information lines. Information on a
bus is arranged so that bit position 7 carries the least
significant bit within an eight-bit byte. The most significant bit is in position 0 and the intervening bits are in
ascending order from position 6 to 1,

TAG IN LINES'
ADDRESS IN
STATUS IN
SERVICE IN
DATA IN

TAG OUT LINES


ADDRESS OUT
COMMAND OUT
SERVICE OUT
DATA OUT

CHANNEL

groups, made up of both inbound (to the CPU) and


outbound (to the SCU) lines, are shown in Figure 3-1.

BUS IN LINES:
NINE BIT POSITIONS

-I

SPECIAL CONTROLS IN
METERING IN
MARK IN 0

SPECIAL CONTROLS OUT:


SUPPRESS OUT
CLOCK OUT
METERING OUT

-\

vi

Figure 3-1. Channel Interface Signal Groups

3-1

3.2 CHANNEL INTERFACE LINES


DESCRIPTION
3.2.1 Inbound Lines
Bus In
The Bus In lines transfer data, address, status, and sense
.information to the channel. The inbound tag lines
indicate the type of information transferred over the Bus
In lines. Either the tag and interlock lines, or scan
controls determine the period during which signals are
valid. The duration of this period is

when OPERATIONAL IN is up,

within 100 nanoseconds after the rise of an


identifying inbound tag to the
responding
outbound tag, or

until SELECT OUT drops in an SCU busy


sequence.

during read and sense operations when information is available on the Bus In lines.
during write and control operations when information is required on the Bus Out lines.

SERVICE IN cannot be up concurrently with any other


inbound tag. When the channel does not respond in time
to the preceding SERVICE IN, an overrun condition
occurs. SERVICE IN does not drop if an outbound tag has
not risen and does not rise if SERVICE OUT has not
dropped.

Status In
Signal STATUS IN informs the channel when the
selected SCU has placed status information on the Bus In
lines. The channel responds by raising SERVICE OUT
(accept) or COMMAND OUT (stack). STATUS IN remains
up until an outbound tag is generated, or until SELECT
OUT drops in an SCU busy sequence. STATUS IN must
drop so that the responding outbound tag may drop. It
cannot be up concurrently with any other inbound tag.
Service In
Signal SERVICE IN informs the channel when the
selected SCUis ready to transmit or receive a byte of
information. It remains up until the channel responds by
raising SERVICE OUT or COMMAND OUT or, during an
interface disconnect sequence, by ADDRESS OUT. The
conditions for generating SERVICE IN are as follows:

when the selection requirements are met by


another channel of a multi-channel SCU.

If the sequence meets the service requirements, REQUEST IN falls not more than 250 nanoseconds after
OPERATIONAL IN goes low. If the request for status
presentation is suppressible, REQUEST IN does not
remain up when SUPPRESS OUT is up. During a
suppressible operation, REQUEST IN falls at the SCU
within 1.5 microseconds after the rise of SUPPRESS OUT
at the SCU.

Data In

Operational In

Signal DATA IN informs the channel when the selected


SCU is ready to transmit or receive a byte of information.
It remains up until the channel responds by raising DATA
OUT or COMMAND OUT or, during aon interface disconnect sequence, by ADDRESS OUT. The conditions for
generating DATA IN are as follows:

Signal OPERATIONAL IN informs the channel that an


SCU has been selected, except during an SCU busy
sequence. OPERATIONAL IN remains up for the duration
of the selection. The address byte transmitted over the
Bus In lines identifies the selected SCU. OPERATIONAL
IN rises only when the inbound SELECT OUT to the SCU
is up and the outbound SELECT OUT is down. The
selected SCU blocks the. transfer of SELECT OUT to the
next SCU. OPERATIONAL IN drops only after SELECT
OUT drops.

during read and sense operations when information is available on the Bus In lines.

during write and control operations when information is required on the Bus Out lines.

Address In
Signal ADDRESS IN informs the channel when the
address of the currently selected SCU and disc drive has
been placed on the Bus In lines. The channel responds by
raising COMMAND OUT. The ADDRESS IN signal
remains up until COMMAND OUT is generated and must
drop so that COMMAND OUT may drop. It cannot be up
concurrently with any other inbound tag.

DATA IN cannot be up concurrently with any other


inbound tag; When the channel does not respond in time
to the preceding DATA IN, an overrun condition occurs.
DATA IN does not drop if an outbound tag has not risen
and does not rise if DATA OUT has not dropped.

When OPERATIONAL IN is raised, it remains up until all


required information is transmitted between the channel
and the SCU. If SELECT OUT is down, OPERATIONAL IN
drops after the rise of the outbound tag associated with
the transfer of the last byte of information. With the
exception of REQUEST IN or METERING IN, all inbound
signals are reset within 1.5 microseconds of the fall of
OPERATIONAL IN at the SCU.

Select In
Signal SELECT IN extends the SELECT OUT signal from
the jumper in the terminator block to the channel. It
provides a return path to the channel for the SELECT
OUT signal.
Request In
Signal REQUEST IN indicates that the SCU is requesting
a selection sequence to present status information.
When OPERATIONAL IN rises, REQUEST IN is dropped
unless

additional selection sequences are required,

when the SCU is no longer ready to present the


status information, or

Metering In
Signal METERING IN is transmitted over a line from all
attached SCUs and is used to condition the CPU meter
for operation. METERING IN rises and falls concurrently
with OPERATIONAL IN for any interface signaling sequence. METERING IN is not raised during the following
conditions:

between generation and acceptance of Device


End,

between generation of Device End and acceptance of the next command during chaining,

while a device is waiting for initiation of an


automatic start.

Mark In 0
Signal MARK IN 0 is used as a marker tag to indicate that
the SCU is requesting command retry. Associated with
MARK IN 0 is the retry status. MARK IN 0 is up when
OPERATIONAL IN is up. Retry status is associated with
MARK IN O.

3.2.2 Outbound Lines


Bus Out
The Bus Out lines transfer data, address, and commands
to the SCUs. The outbound tag lines indicate the type of
information transferred over the Bus Out lines. Either the
tag and interlock lines, or scan controls determine the
period during which signals are valid. The duration of
this period is

from the rise of ADDRESS OUT to the rise of


OPERATIONAL IN or SELECT IN during
transmission of the SCU and disc drive address
signal,

until STATUS IN drops in an SCU busy sequence, or

from the rise of the identifying outbound tag to


the fall of the responding inbound tag during the
transmission of any other type of information.

Address Out
Signal ADDRESS OUT informs the SCU to decode the
SCU and disc drive address on the Bus Out lines. Except
in an SCU busy sequence, the SCU responds when
recognizing the address by raising OPERATIONAL IN
when SELECT OUT rises with ADDRESS OUT still up.
ADDRESS OUT rises not less than 250 nanoseconds
after the SCU and disc drive address have been placed on
the Bus Out lines. ADDRESS OUT must be down for not
less than 250 nanoseconds before being raised for disc
drive selection. Selection of the SCU is cancelled if
ADDRESS OUT drops before SELECT OUT rises.
Except for a disconnect sequence, ADDRESS OUT must
rise only when SELECT OUT, HOLD OUT, SELECT IN,
STATUS IN, and OPERATIONAL IN are down. If HOLD
OUT is down and ADDRESS OUT rises, or ADDRESS
OUT is up and HOLD OUT drops, the SCU will drop its
OPERATIONAL IN and thereby disconnect from the
interface. ADDRESS OUT must remain up until

3-2

OPERATIONAL IN drops. This occurs within 6


microseconds after the disconnect indication (this sequence occurs during a Halt 110 instruction). Mechanical
motion in process continues to a normal stopping point.
Status information is generated and presented to the
channel when appropriate; in this case, ADDRESS OUT
may be up concurrently with another outbound tag.

When SERVICE OUT is transmitted in response to


SERVICE IN during a Read or Sense operation, or to
STATUS IN, SERVICE OUT must rise after the channel
accepts the information on the Bus In lines. During these
operations, SERVICE OUT indicates that the information
on the Bus In lines is no longer required to be valid and is
not associated with any information on the Bus Out
lines.

Command Out

When SERVICE OUT is transmitted in response to


SERVICE IN during a Write or Control operation, SERVICE OUT indicates that the channel has provided the
requested information on the Bus Out lines. In this
operation, SERVICE OUT rises after the information has
been placed on the Bus Out lines. SERVICE OUT remains
up until the fall of SERVICE IN. SERVICE OUT cannot be
up concurrently with any other outbound tag, except
during an interface disconnect sequence (when ADDRESS OUT may be up).

Signal COMMAND OUT informs the selected SCU in


response to a signal on ADDRESS IN, SERVICE IN, or
STATUS IN. During initial selection sequence, raising
COMMAND OUT in response to ADDRESS IN indicates to
the selected SCU that the channel has placed a command byte on the Bus Out lines (the command byte has a
fixed format). The rise of COMMAND OUT indicates that
information on the Bus In lines is no longer required to
be valid. COMMAND OUT stays up until the fall of the
associated ADDRESS IN, STATUS IN, or SERVICE IN tag.
COMMAND OUT cannot be up concurrently with any
other outbound tag line, except possibly during an
interface disconnect sequence For this case, ADDRESS
OUT may be up. DUring a channel-initiated selection
sequence, COMMAND OUT indicates that BUS OUT
defines the operational command to be performed.
During an SCU-initiated sequence, COMMAND OUT is
interpreted as follows

Proceed with sequence if in response to ADDRESS IN.

Stop sequence if in response to SERVICE IN or


DATA IN.

Stack sequence if in response to STATUS IN.

Service Out
Signal SERVICE OUT IS issued to the selected SCU in
response to a SERVICE IN or STATUS IN signal from the
SCU. SERVICE OUT indicates to the selected SCU that
the channel has accepted the information on the Bus In
lines or has provided the data requested by SERVICE IN
on the Bus Out lines.

When SERVICE OUT is transmitted to the SCU in


response to STATUS IN while SUPPRESS OUT is up,
SERVICE OUT indicates to the SCU that the operation is
being chained and that this status is accepted by the
channel SERVICE OUT remains up until the fall of
STATUS IN
Data Out
Signal DATA OUT, transmitted over a tag line from the
channel to all attached SCUs, is issued in response to
DATA IN from the selected SCU. DATA OUT indicates
that the channel has accepted the information on BUS IN
or has provided the data requested by DATA IN on Bus
Out lines.
Select Out
Signal SELECT OUT is sent on an outbound tag line
connected serially through each SCU and is used to
select or poll an SCU. An SCU is selected by raising
SELECT OUT and ADDRESS OUT; polling an SCU is done
by raising SELECT OUT without ADDRESS OUT. SELECT
OUT and SELECT IN form a loop from the channel
through each SCU to the cable terminator block (SELECT
OUT), and through each SCU back to the channel
(SELECT IN). SCU selection cirCUitry can be attached to
either SELECT OUT or SELECT IN.

The logical operation of the SCU ensures that the


process of logically bypassing SELECT OUT before power
changes does not interfere with the propagation of
SELECT OUT. By USing a special latch circuit. the SCU
ensures that SELECT OUT discontinuities that may occur
when the SCU is powered up do not affect propagation of
SELECT OUT. If the SCU is powered up, the latch is
turned on by the ANDing of SELECT OUT and HOLD OUT
and is reset by HOLD OUT going low, The latch circuit is
in series with the remaining selection circuitry in the
SCU and provides a constant SELECT OUT within the
SCU and to the other SCUs, regardless of variations in
the incoming SELECT OUT signal.
The following description pertains to SCU operation with
a properly operating latch circuit. The SCU raises its
OPERATIONAL IN only when its incoming SELECT OUT
goes high. If the SCU does not require selection, it
propagates the signal to the next SCU. When the SCU
propagates SELECT OUT, it does not raise OPERATIONAL
IN or respond with an SCU busy sequence until the next
rise of SELECT OUT. When an operation is being initiated
by the channel. SELECT OUT rises a minimum of 400
nanoseconds after ADDRESS OUT goes high (the address of the SCU and disc drive being selected). The
channel maintains SELECT OUT up until either SELECT
IN, ADDRESS IN and OPERATIONAL IN, or STATUS IN
rises
The SCU becomes selected when it raises
OPERATIONAL IN. SELECT OUT drops to allow
OPERATIONAL IN to drop, but after SELECT OUT is
dropped the SCU maintains OPERATIONAL IN up until
the associated signal sequence is completed. If the SCU
raises OPERATIONAL IN, it prevents SELECT OUT from
being propagated to the next SCU. If the SCU is not
selected, it propagates SELECT OUT within 1.8
microseconds to the next SCU.
When STATUS IN rises in response to SELECT OUT
during a selection sequence (indicating that the SCU is
busy). SELECT OUT drops and does not rise until
ADDRESS OUT has been reset.
Hold Out
Signal HOLD OUT is used in conjunction with SELECT
OUT to provide synchronization of SCU selection. HOLD

OUT is also used to minimize propagation effects when


SELECT OUT goes low by purging the SELECT OUT
signal from the SELECT OUT signal path. When HOLD
OUT drops, it does not rise again until at least 4
microseconds. The minimum down time of HOLD OUT is
2 microseconds under any other condition.
Operational Out
Signal OPERATIONAL OUT connects the channel to all
SCUs and is used for interlocking purposes. All lines
from
the
channel
are
significant
only when
OPERATIONAL OUT is up, except for SUPPRESS OUT.
When OPERATIONAL OUT is down, all inbound lines
from the SCU must drop and any operation currently in
process over the interface must be reset. Under these
conditions, all SCU-generated interface signals go down
within 1.5 microseconds after the fall of OPERATIONAL
OUT at the SCU.
Suppress Out
Signal SUPPRESS OUT is transmitted over a line from
the channel to all attached SCUs. It is used both alone
and in conjunction with the outbound tag lines to provide
the following special functions: suppress data, suppress
status, command chaining, and selective reset. (Additional information about the use of SUPPRESS OUT
during status and data suppression is contained in
paragraphs 3.5.6 and 3.5.12.)
Clock Out
Signal CLOCK OUT is sent out from the channel to all
attached SCUs and is used to provide the CPU interlock
control necessary for changing the enable-disable states
of SCUs. The CLOCK OUT signal must be down to permit
changing enable-disable states. The minimum duration
of the CLOCK OUT down state is 1 microsecond.

Metering Out
Signal METERING OUT is transmitted over a line from
the channel to all attached SCUs. It is used to condition
all other meters in the SCUs and disc drive modules. It is
raised whenever the CPU customer meter is recording
time. (The 3672 SCU does not use METERING OUT.)

3-3

3.3 INTERFACE SEQUENCES


3.3.1 Overall Descriptions
3.3.1.1 INITIAL SELECTION SEQUENCE
The channel places the address of the desired SCU,
controller, and the disc drive on the Bus Out lines and
raises ADDRESS OUT (Figure 3-2). Each SCU connected
to the channel attempts to decode the SCU address on
the bus and see if it compares with the unique address
assigned each SCU. To be accepted by the addressed
SCU, the address must have correct parity.
The channel then issues SELECT OUT. The SCU blocks
propagation of SELECT OUT and raises OPERATIONAL
IN. When OPERATIONAL IN rises, the channel responds
by dropping ADDRESS OUT. After ADDRESS OUT falls
and the SCU, controller, and disc drive address is on the
Bus In lines, ADDRESS IN may rise. After the channel
confirms the address. it responds by placing the command on the Bus Out lines and raising COMMAND OUT.
The selected SCU processes the command and drops
ADDRESS IN, which allows COMMAND OUT to fall. After
COMMAND OUT drops, the SCU places the status
information on BUS IN and raises STATUS IN.

Write and Control operations, DATA IN rises when data


is required on the Bus Out lines. When DATA IN is
alternated with SERVICE IN, DATA IN may rise when
SERVICE OUT is raised in response to SERVICE IN.
However, DATA IN is not considered valid until SERVICE
IN is dropped. Similarly, SERVICE IN may rise when
DATA OUT is raised in response to DATA IN. However,
SERVICE IN is not considered valid until DATA IN is
dropped.
When DATA OUT is sent in response to DATA IN during
a Sense or Read operation, the DATA OUT signal must
rise after the channel accepts the information on the Bus
In lines. In these cases, DATA OUT indicates that the
information is no longer required to be valid on the Bus
In lines and is not associated with any information on the
Bus Out lines. When DATA OUT is sent in response to
DATA IN or a Control or Write operation, DATA OUT
indicates that the channel has provided the requested
information on the Bus Out lines. DATA OUT must
remain up until DATA IN goes low.
Detailed flow diagram descriptions of the read and write
data transfers are presented in paragraphs 3.8 and 3.9,
respectively.
3.3.1.3 ENDING SEQUENCE

If the channel accepts this status condition, it responds


with SERVICE OUT. SERVICE OUT allows STATUS IN to
fall, completing the initial selection sequence.
If the command is rejected by the SCU, the SCU presents
the Unit Check status condition. No operation is initiated
at the SCU and no ending status is generated.
3.3.1.2 DATA TRANSMISSION SEQUENCE
Data transfers may be initiated by the selected SCU aft~r
the selection sequence. To transmit data to the channel,
the SCU places a data byte on the Bus In lines and raises
DATA IN (Figure 3-2). The tag and validity of the Bus In
lines must be maintained until the outbound tag is raised
in response.
To request data from the channel, DATA IN is raised, and
the channel places the data on the Bus Out lines and
signals with DATA OUT. The channel maintains validity
of the Bus Out lines until DATA IN falls. After DATA IN
falls, the channel responds by dropping DATA OUT.
During Sense and Read operations, DATA IN rises when
data is' available on the Bus In lines. During Control and

The ending sequence may be initiated by either the


channel or SCu. If initiated by the channel. the disc drive
may still require time to reach the point where the proper
status information is available, in which case a second
signal sequence is necessary to complete the ending
sequence. If by the SCU, the ending sequence is
completed in one Signal sequence, assuming that both
the Channel End (CE) and Device End (DE) status
conditions occur together (Figure 3-2).
Upon initiation of the ending sequence, one of three
situations may exist' (assume selection is already obtained):

1. The channel recognizes the end of an operation


before the disc storage subsystem reaches its
ending point. For this case, the SCU raises
SERVICE IN whenever the SCU requires service
again. The channel responds with COMMAND
OUT, indicating stop. The SCU drops SERVICE
IN and proceeds to its normal ending point
without requesting further service. When the
subsystem reaches the point at which it would
normally send CE, the SCU places the ending

status on the Bus In lines and raises STATUS


IN. The channel responds with SERVICE OUT,
unless it is necessary to stack the status
condition, in which case the channel responds
with COMMAND OUT.

2. The channel and subsystem recognize the end


of an operation simultaneously. Status information is available at the SCU. The SCU places the
ending status on the Bus In lines and raises
STATUS IN.
3. The subsystem recognizes the end of an operation before the channel reaches the end of an
operation. For this case also, status information
is available at the SCU. The SCU places the
ending status on the Bus In lines and raises
STATUS IN.

raises both ADDRESS IN and OPERATIONAL IN. When


the channel recognizes the address, COMMAND OUT is
sent to the SCU, indicating Proceed. After ADDRESS IN
drops, the channel responds by dropping COMMAND
OUT.
3.3.1.6 IMMEDIATE COMMAND SEQUENCE
An immediate command is one whose execution meets
the following requirements:

Execution requires no more information than


that in the command byte; that is, no data or
information bytes are transferred.

Channel end time coincides with initial status


time and, on a normal operation, at least
Channel End instead of Zero status will be in
the initial status byte.

If DE does not occur with CE, DE is presented when it is


available and an additional status sequence is required.
Additional information pertaining to the stop condition is
presented in paragraph 3.5.11.

A channel response of COMMAND OUT to STATUS IN


cannot prevent the execution of an immediate command.

3.3.1.4 SCU BUSY RESPONSE

3.3.2 Initial Selection Sequence

The SCU busy response begins when the channel places


the 1/0 address on the Bus Out lines and raises
ADDRESS OUT. SELECT OUT is then raised. Each SCU
attempts to decode the address on the Bus Out lines.
When SELECT OUT rises at the addressed SCU, the SCU
blocks propagation of SELECT OUT, places the Busy
status byte on BUS IN, and raises STATUS IN.
OPERATIONAL IN is not raised.

The following two paragraphs describe the initial selection of an SCU, controller, and disc drive in terms of the
operations performed jointly by the channel and SCU to
effect the selection. Paragraph 3.3.2.1 is a simplified
description of the sequence, discussing' operation of the
channel and SCU only. Paragraph 3.3.2.2 discusses the
initial selection in greater detail by including channel
interface and microprogram operations.

After accepting the status byte, the channel drops


SELECT OUT. The SCU responds by dropping STATUS IN
and disconnecting from the interface. The channel must
keep ADDRESS OUT up until STATUS IN drops, thereby
completing the SCU busy response.
The SCU busy sequence is not used in response to an
initial selection sequence addressed to a disc drive for
which chaining has just been indicated.
3.3.1.5 SCU-INITIATED SEQUENCE
An SCU requiring service initiates a service sequence by
raising REQUEST IN to the channel. The next time
SELECT OUT rises at any SCU requiring service and no
1/0 selecting sequence is being attempted by the
channel (ADDRESS OUT down), the SCU places the
address of the SCU and disc drive on the Bus In lines and

3.3.2.1 SIMPLIFIED SEQUENCE


The channel sends SCU and the address of the requested
disc drive to all attached SCUs. In order of priority, each
SCU compares its address to the address sent by the
channel. If the addresses compare (are equal), the
addressed SCU sends its address and the address of the
requested disc drive to the channel.
The channel then checks the address received with
address sent. If the addresses do not compare,
channel turns on Interface Control Check. If the
dresses compare, the channel sends a command to
addressed SCU.

the
the
adthe

The SCU accepts the command and sends its status to


the channel. If the status is to be accepted, the channel
responds with SERVICE OUT which tells the SCU it is

3-4

\It't H\lllN\l lllll

\11Ilfll SS ,lUI

HlllO tlUI

SflEl'10UT

--::S

REQUEST IN

---:5

SELECT IN

--5

OPERATIONAL IN

ADDRESS IN

COMMAND OUT

STATUS IN

---s

SERVICE OUT

--S

I
SERVICE IN

~--~r----~f5~1----+-------~>

I
I

,5 I

BUS OUT 19 LINES)

CEo DE STATUS

~--~'~----~5~

BUS IN (9 LlNESI

~~----------~

SUPPRESS OUT

--4~5----------------------~5)'~----~~'---~

"mAC

"'''n''~

SUPPRESS DATA TRANSFER

55 L '

STOP

L !~~~~~ED

STATUS PRESENTED
NOTE

DELAY

..

DATA TRANSMISSION
NORMAL

INPUT

OPERAT(ON

(CHANNEL

-------------t~~1

DELAY

ENDING SEQUENCE

FORCED, SELECTOR MODEl ----------------------~

THE@IS USED TO INDICATE

CHE~KING

OF SIGNAL LEVEL AND/OR BUS CONTENT


BEFORE PROCEEDING

Figure 3-2. Selector Channel Operation Data Transfer Sequences

3-5

If zero: responds with SERVICE OUT to


indicate to selected SCU that channel
has accepted the status.

ready to begin the data transfer. If status is non-zero, the


channel drops SELECT OUT and HOLD OUT and the
sequence is terminated.
The above sequence is implemented by channel- and
SCU-initiated events as listed below, where (C) indicates
channel.

(C)

Raises OPERATIONAL OUT, ADDRESS OUT,


HOLD OUT, and SELECT OUT. Places address
of requested disc drive on BUS OUT.
(SCU)

3.3.2.2 DETAILED SEQUENCE

(C)

Compares address on BUS OUT


with address of SCU.

(CI)

If address compares, and SCU


is not busy: raises OPERATIONAL IN.

(C)

(M)

Recognizes fall of ADDRESS OUT.


Places SCU, controller, and disc
drive address on BUS IN. Raises
ADDRESS IN.

Compares address on BUS IN with requested


address sent to SCU.
If address does not compare: sets Interface Control Check.
.

(CI)

If address compares: places command on


BUS OUT and raises COMMAND OUT.

(C)

Places disc drive status on BUS IN.


Raises STATUS IN.

(CI)

(C)

IC)

Reset CHF Register bit 3 (Address


Out). Gate SCU, controller, and disc
drive address from BI Register to BUS
IN and raise ADDRESS IN.

Channel drops SERVICE OUT.


(CI)

Reset CHF" Register bit 7 (Service Out).

(CI)

Gate contents of BI Register to BUS


IN. Raise STATUS IN to channel with
SCU status on BUS IN.

Channel checks for zero status. If status is


non-zero, channel drops SELECT OUT and
HOLD OUT. If status is zero, channel raises
SERVICE OUT.
(CI)

Set CHF Register bit 7 (Service Out).


(M)

Microprogram recognizes resetting of Service Out bit.


Microprogram decodes and
executes command and if
command
requires
it,
prepares for data transfer.

3.3.3 Ending Sequence


The following two paragraphs describe the ending sequence of an SCU, controller, and disc drive in terms of
the operations performed jointly by the channel and SCU
to effect the ending sequence. Paragraph 3.3.3.1 is a
simplified description of the sequence, discussing operati~n ,of the channel and SCU only. Paragraph 3.3.3.2
~Iscus.ses the,ending sequence in greater detail by
mcludl.ng
channel
interface
and
microprogram
operations.

3.3.3.1 SIMPLIFIED SEQUENCE


(C)

Microprogram recognizes resetting of Command Out bit in


CHF Register, determines
SCU status, places status in
BI Register and sets CT
Register bit 5 (Status In).

Set CT Register bit 5 (Status In). BI


Register contains SCU Status.

Channel recognizes OPERATIONAL IN and


drops ADDRESS OUT.
(CI)

Reset CHF Register bit 4 (Command


Out).

(CI)

Checks for zero status.


If non-zero: channel drops SELECT OUT
and HOLD OUT and the sequence is
terminated.

Reset CT Register bit 4 (Address In).


Drop ADDRESS IN to channel.

(M)

(C)

Microprogram recognizes setting of Command Out bit in


CHF Register and resets CT
Register bit 4 (Address In).

Channel recognizes the fall of ADDRESS IN


and drops COMMAND OUT.
(CI)

Microprogram recognizes INT


5 branch condition, sets bit 3
(Operational In) and bit 4 (Address In) in Channel Tag (CT)
Register, and places SCU,
controller, and disc drive address in Bus In (BI) Register.
Microprogram records address from BO Register for
later use.

Set bit 3 (Operational In) and bit 4


(Address In) of CT Register. BI Register
contains SCU, controller, and disc
drive address. Raise OPERATIONAL
IN.

Generate Set BO which causes the


command on BUS OUT to be loaded
into BO Register. COMMAND OUT
from channel sets bit 4 (Command
Out) in CHF Register. 80 Register
contains command.
(M)

* When SCU goes not-busy, SCU initiates a POlling Interrupt. This


places SCU End in the Status Byte.

Recognizes OPERATIONAL IN and drops ADDRESS OUT.


,

(SCU)

Compare SCU address to address on


BUS OUT. If equal, set SCU Selected
Latch. Set Switched to A, B, C, or D
Latch. If SCU is busy, SCU responds to
channel with STATUS IN. If SCU is not
busy,* set Interrupt (INT) Register bit 5
(Channel Interrupt). Set Channel Flags
(CHF) Register bit 3 (Address Out). Bus
Out (BO) Register contains address.

(C)

(M)

Raise OPERATIONAL OUT, ADDRESS OUT,


HOLD OUT, and SELECT OUT. Place requested SCU address on BUS OUT.

If address compares, but SCU


is busy: raises STATUS IN.

(SCU)

Channel compares address on BUS IN with


address sent to SCU to ensure that they are
the same. Place Command (Read, Write, etc.,)
on BUS OUT and raise COMMAND OUT.
(CI)

The following list of events provides a detailed description of the initial selection sequence. Abbreviations are
as. follows: channel (C), channel interface (CI), and
microprogram (M). Registers listed are described in detail
in Section 6.

If address does not compare:


sends SELECT OUT/SELECT
IN to next lower priority SCU.

(C)

(C)

Channel word count equals zero. Channel


responds to DATA IN or SERVICE IN with COMMAND OUT.

-OR-*
(SC,U)

(C)

Channel drops SELECT OUT, HOLD OUT, and


raises SERVICE OUT.
(SCU)

(C)

SCU recognizes COMMAND OUT to


stop data transfer. When word count
from track field equals 0, SCU places
status on BUS IN and raises STATUS
IN.

SCU recognizes SERVICE OUT and


drops STATUS IN and OPERATIONAL
IN.

Channel drops SERVICE OUT (chaining is indicated by SUPPRESS OUT being up).
(SCU)

Microprogram recognizes setting of Service Out bit and


drops CT Register bit 5
(Status In).

Recognizes fall of SERVICE OUT and


drops OPERATIONAL IN. If SCU has
additiona I status to present, it is
presented via polling sequence.

* This sequence begins with either the (C) or (SCU)


(CI)

Drop CT Register bit 5 (Status In).

operation.

3-6

3.3.3.2 DETAILED SEQUENCE


The following list of events provides a detailed description of the ending sequence. Abbreviations are as
follows: channel (C). channel interface (CI). and
microprogram (M). Registers listed are described in detail
in Section 6.

(CI)
(C)

(SCU)

Drop STATUS IN to channel.

Channel drops SERVICE OUT (chaining is


indicated by SUPPRESS OUT).

(C)
(SCU)

(CI)

SCU drops ADDRESS IN.

Reset CHF Register bit 7 (Service


Out).

(CI)

Microprogram recognizes setting of bit 5 of B(lNT) branch


condition and stops read or
write data transfer.

(CI)

Reset Channel Control (CHC) Register


bit 0 (Channel Transfer).

(CI)

Drop DATA IN or SERVICE IN.

Drop COMMAND OUT


(M)

When word count from count


field of the record equals zero.
microprogram sets SCU
status into Bus In (BI) Register
and sets Channel Tag (CT)
Register bit 5 (Status In).

(CI)

Set CT Register bit 5 (Status In).

(CI)

Gate BI Register onto BUS IN. and


raise STATUS IN to channel with
status on BUS IN.

(C)

(CI)

Drop OPERATIONAL IN to channel.

Microprogram recognizes setting of Service Out bit and


resets CT Register bit 5
(Status In)

Reset CT Register bit 5 (Status In).

(CI)

SELECT OUT and

(CI)

(M)

Set CT Register bit 0 (Request In).


Raise REQUEST IN to channel.
(C)

(CI)

and

(CI)

Drop REQUEST IN and raise


OPERATIONAL IN and ADDRESS IN.
Place SCU, controller, and disc drive
address from 81 Register on BUS IN

recognizes resetting of Complaces SCU, controller, or disc


in BI register, and sets CT
(Status In).

(CI)

Gate BI Register to BUS IN. and raise


STATUS IN with SCU. controller. or
disc drive status on BUS IN.
Channel stores SCU. controller. or disc drive status
and responds with SERVICE
OUT. Channel drops SELECT
OUT and HOLD OUT.

Set CHF Register bit 7 (Service Out).

Microprogram recognizes setting of Service


Ott bit and resets CT Register bit 5 (Status In)
(CI)

Reset CT Register bit 5 (Status In)

(CI)

Drop STATUS IN
(C)

(M)

COMMAND

Set CT Register bit 5 (Status In).

Prevent propagation of SELECT OUT.


Set SELECT OUT latch and B(lNT)
branch condition.

Set CT Register bit 4 (Address In) and


bit 3 (Operational In). Reset CT
Register bit 0 (Request In).

drops

(CI)

(C)

(M)

Channel
OUT.

Reset CHF Register bit 4 (Command


Out).

Microprogram
mand Out bit.
drive status
Register bit 5

Channel recognizes REQUEST


IN and raises SELECT OUT
and HOLD OUT.

Microprogram recognizes B(lND branch condition, resets CT Register bit 0 (Request In).
and sets CT Register bit 4 (Address In) and
bit 3 (Operational In). Place appropriate
device address in Bus In (BI) Register
(CI)

SCU prevents propagation of SELECT OUT;


drops REQUEST IN; raises OPERATIONAL IN
and ADDRESS IN; and places SCU. controller.
and disc drive address on BUS IN
Channel saves SCU address
raises COMMAND OUT

Reset CT Register bit 4 (Address In).


Drop ADDRESS IN.

Channel drops SERVICE OUT.

SCU. controller. or disc drive has status to


present to channel. Microprogram sets
Channel Tag (CT) Register bit 0 (Request In).
(CI)

(M)

SCU has status to present to channel and


raises REQUEST IN.

(C)

Microprogram recognizes setting of Command Out bit and resets CT Register bit 4
(Address In).

(CI)

3.3.4.1 SIMPLIFIED SEQUENCE

Channel raises
HOLD OUT.

(M)

SCU drops STATUS IN and OPERATIONAL IN.

The following list of events provides a detailed description of the polling sequence and status presentation.
Abbreviations are as follows: channel (C). channel
interface (CI). and microprogram (M). Registers listed are
described in Section 6.

The following two paragraphs describe the SCU-initiated


polling sequence and status presentation of a subsystem
10 terms of the operations performed jointly by the
channel and SCU to effect the sequence. Paragraph
3.3.4.1 is a simplified description of the sequence.
discussing operation of the channel and SCU only.
Paragraph 3.3.4.2 discusses the polling sequence in
greater detail by including the channel interface and
microprogram operations.

(C)

Channel stores status. raises SERVICE OUT. and drops SELECT OUT
and HOLD OUT.
.

Channel saves SCU, controller, or disc drive address


and raises COMMAND OUT.

Set Channel Flags (CHF) Register bit 4


(Command Out).

(C)

(M)

3.3.4 Polling Sequence and


Status Presentation

(SCU)

(CI)

If SCU has additional status to


present. it is presented via
polling sequence.

Set CHF Register bit 7 (Service Out).


(M)

(SCU)

(CI)

3.3.4.2 DETAILED SEQUENCE

OPERATIONAL IN to channel is dropped.


(M)

Channel stores status 10 channel. Channel


drops SELECT OUT. HOLD OUT .. and raises
SERVICE OUT
(CI)

Reset CT Register bit 3 (Operational


In).

(C)

(SCU)
(C)

Microprogram resets CT
Register bit 3 (Operational In).

(CI)
Truncation will set B(lNT) branch condition in microprocessor.
(M)

(C)

(M)

Channel word count equals zero. Channel


responds to DATA IN or SERVICE IN with
COMMAND OUT.

Channel drops COMMAND OUT

SCU places SCU, controller. or disc drive


status on BUS IN and raises STATUS IN.
(C)

(C)

(C)

Channel drops SERVICE OUT.

Microprogram drops OPERATIONAL IN.


(CI)

Reset CT Register bit 3 (Operational


In). Drop OPERATIONAL IN.

3-7

3.4 CHANNEL OPERATION


The SCU IS capable of communicating on two different
channel modes selector mode and block multiplexer
mode Operating characteristics of the SCU when com
munlcatlng ", either of these modes are discussed
below

3.4.1 Selector Mode Operation


The normal mode of SCU operation is in a channel-forced
burst mode. The channel is a block multiplexer channel
operating in a selector mode. The mode is implemented
bv setting a channel mode bit in a CPU control register.
This bit is set to enable the selector mode at IPL or upon
system reset, and can be altered by programming at any
time. The burst mode indicates that the complete block of
data called for by the CCW from one device is transferred
as a complete entity, with no interleaving of data called
for by other CCWs associated with other devices. The
burst mode is implemented by holding up OPERATIONAL
IN for the d.uration of the data transfer
Individual sequences of the selector mode manner of
operation have been described In paragraphs 3 3 1 and
3.3.2. They are shown In terms of the Interface slgnal&
raised and lowered on Figures 3-2 and 3-3

3.4.2 Block Multiplexer Operation


3.4.2.1 INTRODUCTION
The block multiplexer channel enables data to be
transferred between the CPU and storage devices at
higher effective rates of speed than is possible with the
selector channel. The higher effective data rates allow
attachment of high-speed liD devices such as the 3672based subsystem.
A block multiplexer channel performs differently from a
selector channel in the way that command-chained
channel programs are handled. While executing such
programs, a selector channel or a block multiplexer
channel operating in a selector mode, is busy during the
entire time the channel program is in operation, whether
data transfer is occurring or not. A block multiplexer
channel executing a command-chained channel program
has the ability to disconnect from the operational
channel program during certain non-data transfer
operations. Such operations as disc drive seeking are
channel nonproductive activities. A block multiplexer
channel can be freed during a nonproductive activity to

allow more data to be transferred per Unit of channel


busy time
A Single block multiplexer channel can support Inter
leaved
concurrent execution of multiple channel
programs It has multiple subchannels. each of which
has an associated unit control word (UCW) UCWs are
stored In normally unaddressable auxiliary storage of
16K bytes, referred to as "bump" storage and can
support one liD operation
To facilitate channel scheduling, an interrupt condition
called Channel Available has been defined for internal
use on block multiplexer channels. At disconnect time for
a channel program, the channel is available for the
resumption of an uncompleted channel program
previously started, or' another channel program can be
initiated. A Channel Available interrupt occurs at disconnect time to indicate channel availability if a Start
110, Test 110, Test Channel, or Halt 110 instruction was
Issued previously while the block multiplexer channel
was busy
The block multiplexer mode is enhanced through the use
of two standard features of the 3672-based subsystem
rotational position sensing (RPS) and disconnected command chaining. These two features, together with block
multipleXing, increase system throughput by increasing
channel throughput
The use of RPS frees the channel more often during drive
operations. For example, during the time required to
position a track to a specific record, channel programs
are permitted to be initiated sooner on block multiplexer
channels than is possible with selector channels.
Disconnected command chaining is implemented in the
SCU to enable it to handle concurrent execution of
multiple channel programs. For example, the SCU can
simultaneously control many channel programs, one for
each of its drives.
3.4.2.2 ROTATIONAL POSITION SENSING
Use of RPS reduces the time the channel is busy
searching for a disc record. It permits a Search command
to be initiated just before the desired record is to come
under the Read/Write head; that is, when the desired
rotational position is reached. To accomplish this, the
tracks in each cylinder of a drive are considered to
consist of equally spaced sectors. A sector is defined as
the length of track arc that passes under the Read/Write

heads In approximately 130 microseconds Track formatting is unchanged but each record has a sector number
as well as a record address. There are 128 sectors per
track on the 367X drive, with each sector allocated a
unique sector number (00-7F hexadecimal). See Figure
3-4
The SCU can determine the sector currently under the
Read/Write heads of each of its drives. A sector counter
is contained in each drive. The counter is incremented
once every sector time period (approximately 130
microseconds) and set to zero each time the index
marker passes under the heads. The sector in which a
record falls is a function of the length of all records that
precede it and of its sequential position on the track.
Therefore, the sector location can be calculated for fixedlength records and a sector number allocated.
Two disc commands are provided for use with rotational
position sensing: Set Sector and Read Sector. Set Sector
used in conjunction with the block multiplexer channel
permits a single command-chained channel program to
be initiated for each disc operation that frees the channel
and SCU during rotational pOSitioning operations. If the
sector address of a record is known or can be calculated,
a Set Sector command can be included in the disc
channel program to cause the SCU to look for the
designated sector. Once the SCU accepts the sector
number, both the SCU and the channel disconnect are
available for another liD operation.
The Read Sector command is useful for sequential disc
processing and for write verification. When chained from
a Read, Write, or Search command, Read Sector provides
the sector number required to access the record processed by the previous CCW. This sector number can be used
to reposition the track to verify the record just written or
to read or write the next sequential record.
3.4.2.3 DISCONNECTED COMMAND CHAINING
Disconnected Command Chaining (also referred to as
multiple requesting) is used for the following purposes:

Allows the SCU to disconnect from the channel


on commands which require long delays due to
mechanical motion or searches. The channel
and SCU are free during the delay period.

Enables multiple requesting, which allows


separate command chains to be active in the
subsystem, one for each disc drive.

A block diagram of Disconnected Command Chaining is


shown in Figure 3-5.
Disconnected Command Chaining enables the SCU to
disconnect from the channel after an operation such as
Seek or Set Sector has started even though chaining is
indicated. Since the burst mode is not forced during the
execution of Seek or Set Sector commands, the SCU can
disconnect between Channel End and Device End. The
disconnect function reduces the CPU interrupts needed
to overlap channel data transfers with mechanical motion of the drives. The SCU retains the information
necessary to control a disconnected CCW chain for each
drive in the subsystem. In effect, the SCU is capable of
executing
many CCW chains (one per drive)
simultaneously.
During a Seek operation, the SCU attempts to reconnect
after drive mechanical motion is completed. During a Set
Sector operation, the SCU attempts to reconnect when
the desired sector number is detected. In either case, the
channel is made available during access motion and
rotational delay periods. Disconnect is also allowed on
Command Retry procedures.
Disconnected Command Chaining can involve two
channels, or one channel can have several CCW chains
going.
Two examples of Disconnected Command Chaining are
shown in Figure 3-5.

3.4.2.4 MULTIPLEXER CHANNEL OPERATION


This paragraph contains an explanation of block multiplexing operations. References are made to Figure 3-6.
which illustrates the detailed signal sequences for an
SCU-forced, burst mode, multiplexer channel operation.
This is the normal output operation.
Assume a block multiplexer channel is executing a
channel program consisting of mUltiple commandchained CCWs. When Channel End is presented without
a corresponding Device End, the channel disconnects
from the current channel program and becomes available
for an liD operation via another device. The disconnection occurs even though the current channel program is
not complete At disconnect time, the information
necessary to restart the disconnected channel program is
stored in the UCW aSSigned to the active subchannel and
SCU.

3-8

r----------------------------~5>~f--------------------------~il~--------------------~f5~----------------------~S

OPERATIONAL OUT

HOLD OUT

SELECT OUT

REQu~S T

IN

SELECT iN

OPERATIONAL IN

ADDRESS IN

r.OMMAND 0'J'

STATUS IN

SERVICE OUT

SERJICE IN

55

I
ADD~ESS

C4

CTRL CMD

BUS OUT (9 LlNESI


ADDRESS C4

CE STATUS

ILJI

BUS IN 19 CINES,

~
~~

,
I
ADDRESS C4
I

~----------~5'~-----------I--I-------,------------~f
ADDRESS C4

SCU BUSY STATUS

DE STATUS

11

~------------_7fr-~------~

5
SUPPRESS OUT

51

14-----------

IMMEDIATE COMMANDS --------------;,,~._----CONTROL COMMAND INITIATED


1

t+-------------------------------

CHANNEL-INITIATED SEQUENCES

--------,,*.._----- SCU BUSY RESPONSE

:5

DEVICE END
STATUS OCCURRED

------+0...---------- ENDING

-----------------------------------------------r--------

~If

DEVICE END
STAruS ACCEPTED
SEQUENCE

SCU-INITIATED SEOIIFNf"

NOTE
TH01S USED TO INDICATE CHECKING
OF SIGNAL LEVEL AND OR BUS CONTENT
BEFORE PROCEEDING

Figure 3-3 Selector Mode Operation. Miscellaneous Sequences

3-9

----I
I
~

INDEX
Channel A Example
READ/WRITE

127' 0

SCU

1. Issue Seek command to Dnve 1.


2. Disconnect until Oevice End (seek complete)
IS received.

CHANNEL
A

WRITABLE
CONTROL
STORAGE
(WCS)

CCW Cham Example-Drive 1


1 Seek-Locate cylinder and disconnect until
Device End.

2. Set Sector-Locate area on track and disconnect until Device End.


3. Search ID-Locate a record
4. TIC
5 Write Data -Write data field

READ/WRITE
DATA

MICROPROGRAM

It

I
I
I

DRIVE
1

MICROPROCESSOR

TWO-CHANNEL
SWITCH OPTION.

1 Issue Seek command to Drove 6.


2 Disconnect until DeVice End (seek complete)
from Drive 6 IS received
Issue Set Sector command to Drive 4
Disconnects from this chain until Device End
from Drive 4 (sector IS ready) IS received.

..

CHANNEL
INTERFACE

---- -- -- -- -

Channel B Example

~~ CH_A_~_N_EL
___

____

Multiple Requests on Drives 6 and 4

CONTROLLER
INTERFACE

1 Seek-Locate cylinder and disconnect until


Device End.
2 Read Home Address-Read Home Address
field

CONTROLLER

UJ

II

I-----.JL
Controls each operation as it IS received from
the channel. Maintains status of disconnected
command chaIRS in DCC byte in WCS. Brings up
unsuppresslble status for those drives with a bit
In the DeC byte as thelf Device End is received.

r-----------~I'~-------------------------------------.,
CCW Chain Example-Drive 6

If the channel is busy when reconnection is requested,


the SCU must wait until the channel becomes available.
Once mUltiple channel programs have been initiated on
one channel, the interleaving of data transfer operations
is controlled by block-multiplexer-channel hardware and
the SCUs of disc drives operating in the block multiplexing mode.

'I

II

..

Figure 3-4. Disc Track Sector Format

When Device End is presented to signal that the drive is


again ready for the channel, the SCU attempts to regain
use of the channel by signalling REQUEST IN. If the
channel is free, the active channel registers are reloaded
with the information previously saved in the UCW
subchannel assigned to the drive, and the disconnected
channel program is resumed at the appropriate CCW.

Ir-

CCW ChaIR Example-Drive 4

[J
1
I [J
I
~
[J

Perform Seek from channel


A. Control servo motion.
Interrupt microprogram
when seek is complete.

Perform Set Sector from


channel 8. Check for sector
requested
Interrupt
microprogram when sector
is ready.

.J

Perform Seek from channel


B.

Control

servo motion

Interrupt microprogram
when seek is complete.

1 Seek-Locate cyl mder and disconnect until


Device End.
2. Set Sector-Locate record and disconnect
until Device End.
3 Search ID-Locate record.
4. Read Key and Data-Read Key and Data
fields.

Figure 3-5. Disconnect Command Chaining Block Diagram and Examples

3-10

OPERA'IONAI OU1

If

HOLD OUT

SELECT OUT
REQUEST IN

SELECT IN

STATUS IN

SERVICE 'JUT

SERVICE IN
BUS OUT

BUS IN

DATA OUT
DATA IN

MARK IN 0

COMMAND
RETRY
INDICATION

CHAINING INDICATED
BLOCK MULTIPLEX OPERATION
(DATA IN/DATA OUT TRANSFER)

BLOCK MULTIPLEX OPERATION


(RECONNECTION WITH DEVICE END)

Figure 3-6. Block Multiplexer Channel Operation (1 of 2)

3-11

When multiple channel programs are. operating concurrently in the block multiplexing mode, a device can
regain control of the channel only when the channel is
not busy. Therefore, only cyclic devices, such as the
367X Drive with RPS, can disconnect during execution of
a command-chained channel program on a block multiplexer channel and resume ,operation later.

OPERATIONAL OUT

AOORESS OUT

HOLD OUT

Data transfer for concurrently operating devices on a


block multiplexer channel are interleaved on. a firstcome, first-served basis as the desired records become
available. Therefore, devices are serviced in the order in
which their records become available, not necessarily in
the order in which their channel programs are initiated.

SELECT OUT

REQUEST IN

SELECT IN

OPERATIONAL IN
ADDRESS IN

COMMAND OUT
STATUS IN

SERVICE OUT

SERVICE IN

BUS OUT

BUS IN

SUPPRESS OUT
DATA OUT

DATA IN

MARK IN 0

BLOCK MULTIPLEX OPERATION


(RECONNECTION FOR COMMAND RETRY)

Figure 3-6. Block Multiplexer Channel Operation (2 of 2)

3-12

3.5 SEQUENCE CONTROLS


3.5.1 Command Chaining
The 3672-based subsystem permits command chaining,
which is the ability to execute a series of channel
commands as a result of a single I/O instruction.
Command chaining is specified when bit 33 (CC) in the
CCW is set to 1. Upon completing the current CCW with
CC on, the channel fetches a new CCW that specifies the
next 110 operation. This operation is automatically
executed when the subsystem completes the current
operation and Device End status is accepted by the
channel. Upon completing the current CCW, an I/O
interrupt does not occur, and the count. indicating the
amount of transferred data, is not available to the
program. All subsystem programs normally use command chaining. Time is available in the gap between
record areas to execute command chaining functions
See Section 2 for restrictions on the command sequence
within a chain.
Command chaining is indicated if SUPPRESS OUT is up
when SERVICE OUT is raised in response to STATUS IN.
To ensure recognition of command chaining by the SCU,
SUPPRESS OUT must be up at least 250 nanoseconds
before SERVICE OUT rises in response to STATUS IN and
must not fall before STATUS IN
Command chaining is suppressed if an operation is
terminated with either Unit Check or Unit Exception
status conditions noted.

3.5.2 Data Chaining


Data transferred between CPU main storage and the
subsystem may be chained. This data chaining permits
blocks of data to be transferred to or from noncontiguous
areas of main storage. Data chaining is specified by
turning on bit 32 (CD) of the CCW. The new CCW
command code is ignored unless the command code
(CCW bits 0-7) specifies transfer-in-channel (TIC).
Data chaining occurs immediately after the last byte of
data designated by the current CCW has been
transferred to main storage or accepted by the subsystem.
Data chaining takes precedence and command chaining
is ignored if both data chaining and command chaining
are indicated In the CCW

NOTE
Data chaining "capabilities are
dependent on several variable
factors, such as system type, I/O
configuration, and channel
loading.
Because
of
these
dependencies, Read or Write
data chaining within record
areas may cause unpredictable
overruns or chaining checks.
Refer to the FIPs volume if these
conditions are encountered or
suspected.

3. Drive address in the three least significant bits.


The SCU address is determined by the customer when
the subsystem IS initially installed. A drive address is
determined by logical assignments specified by the driveinserted Logical Address Plug. The SCU accepts any
controller/drive address from 0000 to 1111 (00000 to
11111 for 32-spindle configuration). An operation is
terminated with Unit Check status if multiple responses
to an address occur due to hardware failures or duplicate
(modulo-a) Logical Address Plugs on the same controller.

3.5.5 Stack Status


3.5.3 Branching in Channel Programs
Branching in channel programs allows the next CCW to
be read from a storage address other than the address
normally accessed, which is a positions higher than the
current CCW location. This branching is accomplished
either of two ways:

1. If command chaining is specified in a Search


command, and execution of the command
results in a Status Modifier indication (search
satisfied), the channel fetches the next CCW
from a main storage location, sixteen positions
higher than the current CCW.
2. TIC command may be used to modify the
sequence of a chain of commands. The data
address portion of the TIC CCW specifies the
main storage location of the next CCW.
Therefore, the next CCW may be fetched from
any valid main storage location.

3.5.4 Unit Selection and Device Addressing


The I/O addresses of the 3672 SCU, 3673 Controller,
and 367X Drives are designated by an eight-bit binary
number in an I/O instruction. See paragraph 2.3.2.
These addresses consist of three parts:
1" SCU address in the four most significant bits
(three if a 32-spindle configuration).
2. Controller address in the next least significant
bit(s)

Stack Status is indicated by a COMMAND OUT response


from the channel to STATUS IN (paragraph 3.3.1.3). The
Stack Status indication causes status information to be
retained at the SCU, controller. or disc drive until that
status is accepted on a subsequent status cycle with
SERVICE OUT. When Stack Status occurs. the SCU is
disconnected from the channel interface after SELECT
OUT goes down. and COMMAND OUT remains up until
OPERATIONAL IN falls. An attempt by the SCU to initiate
a selection sequence to present the stacked status is
under control of SUPPRESS OUT.

3.5.7 Disconnect In
The DISCONNECT IN Signal enables the SCU to alert thE.
system of a malfunction that is preventing the SCU frolT'
signalling properly over the I/O interface. In response to
DISCONNECT IN, the channel performs a Selective
"Reset.

3.5.8 Interface Disconnect


The SCU recognizes Interface Disconnect when ADDRESS OUT is up, and SELECT OUT and HOLD OUT are
down at least 250 nanoseconds before the completion of
any signal sequence. In this case, ADDRESS OUT may be
up concurrently with another outbound tag. When
OPERATIONAL IN drops, the channel may drop ADDRESS OUT to complete the interface disconnect sequence. ADDRESS OUT must be down at least 250
nanoseconds before a new channel-initiated selection
. sequence may be initiated.
Except for REQUEST IN, the SCU responds to the
Interface Disconnect by removing all Signals from the
interface. The SCU remains busy after it receives an
Interface Disconnect, while performing an operation,
until the Device End status is received by the channel.

3.5.9 Selective Reset

3.5.6 Suppress Status


Whenever the channel is unable to process an immediate request resulting in status conditions. SUPPRESS OUT may be raised. With this line uP. the SCU
may not attempt to initiate a selection sequence to
present suppressible-type status information." Stacked
Status is suppressible. Status that contains Channel End
is not suppressible until stacked, except when an Interface Disconnect is received for that address. Status that
contains Device End is not suppressible when chaining
has been indicated, until it is stacked. Other
asynchronous status conditions may be suppressible at
the option of the SCU without being stacked.
SUPPRESS OUT must be up at least 250 nanoseconds
before SELECT OUT rises at the SCU if suppression of
status is to be ensured. If SUPPRESS OUT rises after a
status sequence has been started, the status sequence
proceeds normally.

The drive presently operating over the channel interface


may be put into a reset condition, termed Selective
Reset. This condition is indicated whenever SUPPRESS
OUT is up and OPERATIONAL OUT drops. As a reSUlt,
OPERATIONAL IN goes low, and the disc drive operating
over the interface and its status are reset. The operation
in process proceeds to a normal stopping point, if
applicable, with no further data transfer. Only the disc
drive operating over the interface is reset. The disc drive
path is in the busy state throughout the Selective Reset
period.
The ready or not-ready state of the SCU is generally not
changed by a Selective Reset. However, when the SCU's
Enable/Disable or Online/Offline switch was changed
before the reset but had not become effective due to the
required inhibiting conditions, the ready or not-ready
state may change if the reset clears those inhibiting
conditions.
Selective Reset is issued only as a result of a timeout by
the channel or a malfunction detected at the channel.

3-13

3.5.10 Command Retry

3.5.12 Stop (or Truncation)

When the command being executed encounters a condition requiring retry. the SCU indicates this requirement
by raising STATUS IN while presenting Unit Check.
Channel End. and Status Modifier (MARK IN 0 being up
for retry status)

Stop is indicated by COMMAND OUT in response to


SERVICE IN. DATA IN, or as a result of an Interface
Disconnect which occurs before the normal Channel End
for the operation in process. It is used to signal the SCU
that the channel IS ending the current operation. On
receipt of the stop indication. the SCU must proceed to
its normal ending point without sending any further
SERVICE IN signals to the channel. The SCU remains
busy until the ending status is presented and accepted by
the channel.

The channel acknowledges the occurrence of command


retry by accepting the status byte containing retry status
and indicating chaining. When Device End is presented
to the channel, it is accepted with chaining indicated and
a normal reselection occurs to reissue the previous
command. A channel indicates refusal to perform a
Command Retry by accepting the status byte containing
retry status without indicating chaining or by stacking
the status byte.
Additional information about Command Retry is conta ined in Paragraph 2.11.

3.5.11 Proceed
At any time other than during a channel-mitiated selection sequence. proceed is indicated whenever COMMAND OUT responds to ADDRESS IN. It causes the SCU
to continue the normal servicing sequences on the
interface.

For 1/0 operations that have relatively long times


between stop and ending status and no time-dependent
chaining requirements, OPERATIONAL IN is dropped on
receipt of the stop indication.

3.5.13 Suppress Data


Input/Output operations whose rate of data transfer can
be adjusted without overrunning are subject to suppression of data transfer by SUPPRESS OUT. A suppression
of data can occur as follows:
Unless the data transfer is contiguous with
initial selection. SUPPRESS OUT is ignored for

the first data byte of any selection sequence.


This means no deselection and reselection
occurs between initial selection and data
transfer

and IS ready for acceptance by the SCU is indicated by


raising SERVICE OUT in response to SERVICE IN, or
DATA OUT in response to DATA IN.

2 To ensure suppression of subsequent data.


SUPPRESS OUT must be up either 250
nanoseconds before the rise of SERVICE OUT or
DATA OUT. or at least 250 nanoseconds before
SERVICE OUT or DATA OUT falls.

Status placed on the Bus In lines which has been


accepted by the channel is indicated by raising SERVICE
OUT in response to STATUS IN.

3. When SUPPRESS OUT is up at the SCU. the


SCU must not raise SERVICE IN or DATA IN for
subsequent suppressible data.

3.5.17 System Reset

3.5.14 Data Acceptance


During a Read or Sense operation, information placed on
the Bus In lines which has been accepted by the channel
is indicated by raising SERVICE OUT in response to
SERVICE IN, or DATA OUT in response to DATA IN.

3.5.15 Data Ready


Durin.g a Write or Control operation. the requested
information which has been placed on the Bus Out lines

3.5.16 Status Acceptance

A System Reset condition is indicated whenever


OPERATIONAL OUT and SUPPRESS OUT are down
concurrently and the disc drive is in the on-line mode.
This condition causes OPERATIONAL IN to fall. and all
SCUs and their attached controllers and disc drives,
along with their status, to be reset. The SCUs remain in a
busy state for the duration of their reset procedure. The
ready or not-ready state of an SCU is not changed by a
System Reset.
To ensure a proper reset, OPERATIONAL OUT and
SUPPRESS OUT must be down concurrently for at least 6
microseconds.

3.6 STATUS CONDITIONS

Designations of the status conditions. and their pOSitions


within the status byte and CSW, are as follows: .

3.6.1 Overview
When requested by the channel, status information
about the subsystem is transferred to the channel by
means of the Test 110 command. This command is
generated by the channel as a result of the Test 110
instruction, or as an internal channel function when the
channel requests status.

DEVICE
STATUS BYTE,
BIT POSITION

2
3
4

The Test 110 command transfers one status byte to the


channel. This byte is placed on the Bus In lines and is
defined when STATUS IN is high. The status byte reports
status conditions about the SCU, controller. and disc
drive whose addresses appeared on the Bus In lines
(with ADDRESS IN up) during the polling or selection
portion of the sequence. In the case of the SCU busy
sequence. when no ADDRESS IN occurs, it is assumed
that the conditions pertain to the SCU.
NOTE
Unless otherwise stated, information on this page pertains to
SCUs attached to only one
channel interface.
Transfer of the status byte occurs in seven situations:

during the initial selection sequence;

to present Channel End at the end of data


transfer;

to present Device End and any associated


conditions to the channel;

to present Control Unit End or Device End to


Signal that the SCU, controller, or disc drive
previously busy. and then interrogated, is now
free;

to present any previously stacked status when


allowed to do so;

to present any externally initiated status to the


channel because of not-ready to ready transition; and

ConclUSion of a Test I/O command

5
6
7

The SCU provides Control Unit End if the SCU was


previously addressed while in the busy state. The busy
state is defined in the next paragraph.

3.6.2.5 DEVICE END

32
33
34
35
36
37
38
39

3.6.2.3 BUSY
PARITY
A TIENTION (NOT USED)
STATUS MODIFIER
CONTROL UNIT END
BUSY
CHANNEL END
DEVICE END
UNIT CHECK
UNIT EXCEPTION

Busy can occur only during a channel-initiated-selection


sequence. It indicates that the disc drive, controller, or
SCU cannot execute the command because of one of the
following reasons:

3.6.2.1 STATUS MODIFIER

A previously initiated operation is being executed. An operation is being executed from the
time initial status is accepted until Device End is
accepted.

Status conditions exist. The condition's accompany the Busy indication. If the Busy indication
pertains to an SCU function, it is accompanied
by Status Modifier.

Unavailable path exists on a string-switch controller.

3.6.2 Description of Status Conditions


The status conditions that are detected by the disc drive,
controller, or SCU and indicated to the channel over the
1/0 interface by the presentation of the status byte, are
described below. Once the conditions are accepted by the
channel, the status byte is not presented again.

Busy is indicated to the Test 110 command only if a


previously initiated operation is still being executed and
no end status is available. It causes command chaining
to be suppressed.

Device End is caused by completing an 1/0 operation at


the disc drive or controller, or by manually changing the
disc drive from the not-ready to ready state. ("Not ready"
means that the disc drive or controller requires operator
intervention to become operationaL) It normally indicates
that the disc drive or controller has completed the
current operation. Each 110 operation can generate only
one Device End signal. The signal is not generated
unless the command is accepted.
Device End is generated either with Channel End or later.
For data transfer operations, the disc drive terminates
the operation when Channel End is generated, and both
Channel End and Device End occur together. For control
. operations, Device End is generated upon completing the
operation at the disc drive. The operation may be
completed either at the time Channel End is generated or
later. Generation of Device End in special cases is
described as follows:

When command chaining occurs, only the


Device End of the last operation of the chain is
normally made available to the program.

When Device End is received in the absence of


any unusual conditions while chaining, it
causes the channel to initiate a new 1/0
operation. If an unusual condition is detected
during the initiation of a chained command, the
chain is terminated without Device End. Sense
information is available defining the unusual
condition.

An 1/0 device shared between more than one


channel path and which generates Device End
because of the disc drive going from a not-ready
to ready state must present Device End to all
attached channels if the Tag switch is in the
TAG position. Device End will be accepted by
only one channel if the switch is in the UNTAG
position.

If a disc drive is addressed while busy, Device


End must be Signalled to the path that initiates
the command when the disc drive becomes not
busy.

Status Modifier is used in three situations:

For operations that do not cause data to be


transferred, Channel End may be provided duro
ing the initial selection sequence.

DESIGNATION

CSW
BIT

3.6.2.2 CONTROL UNIT END

When presented with Busy, to differentiate a


busy SCU from a busy controllerldisc drive
during the initial selection sequence.
When presented with Device End, to recognize
special ending conditions, such as Search
Equal. when the special condition occurs.
Recognition of special ending (or synchronizing)
conditions indicates that the normal sequence
of commands must be modified.
When presented with Unit Check to indicate
that an unusual condition, calling for a retry of
the last channel command, has occurred.

A busy SCU but not-busy controller Idisc drive occurs


when the SCU is required to perform a function that does
not involve the 1/0 interface. or the SCU has status
pendmg for a disc drive other than the one addressed.

3.6.2.4 CHANNEL END


Channel End indicates the portion of any 1/0 operation
involving the transfer of control information or data has
been completed. Each 1/0 operation can cause only one
Channel End signal to be generated. The Signal is not
generated unless the command is accepted. Acceptance
is indicated by the status byte containing either all zeros.
or Channel End and not Busy. The exact time when
Channel End is generated depends on the operation, as
follows:

For operations such as reading, Channel End is


generated when the block has been read.

During normal control operations, Channel End


is usually generated after the control information is transferred to the SCU. Channel End may
be delayed until the completion of short control
operations.

3-15

3.6.2.6 UNIT CHECK

NOTE
The ending interruption condition
can be cleared by Test 110
without generation of Unit Check
if a disc drive becomes not ready
on completion of a command.

Unit Check status indicates that the SCU, controller, or


disc drive has detected an unusual condition that is
described by the information available to a Sense
command. For example'

Indication that an equipment or programming


error has occurred.
The not-ready state of the disc drive or controller has affected the execution of the command.
Exceptional conditions other than the one identified by Unit Exception have occurred.

An error condition causes Unit Check only when it occurs


during execution of a command or some activity
associated with an 1/0 operation. Unless the error
condition pertains to the activity initiated by a command
and is of significance to the program, the condition does
not cause the program to be alerted until after Device
End has been cleared. A malfunction may, however,
cause the disc drive to become not ready.

preserve the sense data and an available disc


drive path until after Device End is accepted.

Unit Check status is presented when the not-ready state


interferes with proper execution of the command, or
when the command, by its nature, tests the state of the
disc drive as follows:

Termination of an operation with Unit Check


status causes command chaining to be suppressed.

If the condition that precludes proper execution


of the operation occurs after execution has
started, Unit Check accompanies Channel End,
Control Unit End, or Device End, depending
upon when the condition was detected. Errors
detected after Device End is cleared may be
indicated by Unit Check with Control Unit End.

When Unit Check appears with Channel End


and without Device End, a shared SCU must

Errors, such as invalid command code parity or


invalid command code, do not cause Unit Check
when the disc drive is working or contains a
pending interruption condition at the time of
selection. In these situations, the SCU responds
by providing Busy and the pending disc drive
interruption condition, if any. The command
code validity is not indicated.
Selection of a disc drive in the not-ready state
does not cause a Unit Check indication when a
Sense command is issued, and whenever an
interruption condition is pending for the addressed disc drive at the SCU.

Unless a command is designed to cause Unit


Check, Unit Check is not indicated if the command is properly executed, even though the disc
drive has become not ready either during, or as
a result of, the operation. Similarly, Unit Check
is not indicated if the command can be executed
with the disc drive not ready.

If, during the initial selection sequence, the SCU


detects that the command cannot be executed,
Unit Check status is presented to the channel. It
appears without Channel End, Control Unit End,
or Device End. This status indicates that no
. action has been taken at the disc drive in
response to the command.

3.6.2.7 UNIT EXCEPTION


Unit Exception indicates that the SCU detected an
unusual condition such as end of file. It has only one
meaning for any particular command. A sense operation
is not required as a response to the acceptance of a Unit
Exception status condition.
Unit Exception conditions can be generated only when
the disc drive is involved with some activity associated
with an 1/0 operation and the condition is of immediate
significance.
If a condition that precludes normal execution of an
operation occurs after an operation is started, Unit
Exception is accompanied by Channel End, Control Unit
End, or Device End, depending on when the condition is
detected. Unit Exception suppresses command chaining.

3-16

3.7 SENSE CONDITIONS


3.7.1 Overview

Under certain conditions, the sense information that


pertains to the last 1/0 operation or other unit action
may be reset by the next command addressed to the
SCU. These conditions are that the Busy bit is not
included in the initial-selection status byte, except where
the command is a Sense I/O or a Test I/O instruction
addressed to the SCU

DESIGNATION

Command Reject
Intervention Required
Bus-Out Parity
Equipment Check
Data Check
Overrun
(not used)
(not used)

Conditions that occur during the execution of an instruction command sequence are reported as channel status
and as sense information. Sense information defines
conditions that occurred in the SCU, controller, and disc
drive that are not defined in the status. This sense
information is .obtained by executing a Sense command.
Information concerning the actual state of the SCU,
controller, and disc drive; and unusual conditions
detected in the last operation, are provided by data
transfer during a sense operation. It may describe
reasons for the Unit Check indication of a status byte
and, for example, may also contain an indication that the
susvstem is in the not-ready state.

BIT

3
4
5
6
7

Command Reject-Bit 0
This bit indicates that the command cannot be performed
for any of several reasons. Typical examples are listed
below:

Command op code, or its information, is invalid.


For example, a Read command was received
which the subsystem is not deSigned to execute
(Read Backward issued to a drive).

The addressed subsystem was in the not-ready


state.

A Diagnostic Write or Diagnostic Load command


was issued, but the microdiagnostic is not
resident in the SCU.

@;

Sequence of commands or the protected conditions has been violated. For example, a Write
command was received which the subsystem
cannot execute because the command violates
the file mask.

A total of 24 bytes of sense information is transferred to


the system by the SCU when it receives a Sense I/O
command. The first two bytes normally provide all sense
information significant to the use of the subsystem. The
bit-by-bit meaning of messages reported in bytes 0
through 6 is preassigned. The meaning of messages
reported in bytes 8 through 23 depends on the format
used for these bytes. This format is decoded in byte 7,
bits 0 to 3.

An additional sense bit may define the reason for the


rejection.

3.7.2 Conditions Indicated by Bits


of Sense Byte 0

Intervention Required-Bit 1

The bits making up sense byte 0 are independent of each


other and are designated as follows:

This bit indicates that the subsystem is either not


connected or for some reason is not ready.

Write portion of the file mask has been violated.


For example, an invalid sequence of commands
was recognized (a Write command is issued to a
drive without previously performing Search).

An uncorrectable data error has been detected


in the information received from a disc driv ~
(condition further defined in sense byte 7).

The bit is set for an error in the count field with a


secondary indication but is not set for an error in the key
field during a read data operation.
The SCU forces correct parity on data sent to the
channel. Data errors on reading and writing cause the
operation to be terminated prematurely only when the
errors prevent meaningful continuation of the operation.

Bus-Out Parity-Bit 2

Overrun-Bit 5

Bus-Out Parity is generated when the SCU detects


invalid parity in a data or command byte received from
the channel. It indicates that incorrect data may have
been recorded during a write operation. However, writing
the incorrect data does not cause the write operation to
be terminated prematurely, unless an error prevents
meaningful continuation of the operation.

This bit indicates that one of the following conditions has


occurred:
1. The SCU received a byte from a disc drive before
the last byte read was accepted by the channel.
2. During a write operation, a data byte was
received too late from the channel.

Equipment Check-Bit 3

The sense information may be changed when not-ready


to ready Device End status is generated, or as a result of
other such asynchronous actions.

3672.21-0001-10175

Intervention Required is generated when the subsystem


is in the test mode or the not-ready state. It is also
generated when the last operation could not be executed
because of a condition that requires intervention at the
subsystem. For example

This bit indicates that an unusual hardware condition,


originated in the channel, SCU, controller, or disc drive
has been detected. The condition is defined by the bits in
sense bytes 7 through 23.
During output operations, the unusual condition or
malfunction may have caused invalid data to be recorded.
When the malfunction prevents any meaningful continuation of the operation, Equipment Check is generated
to stop data transmission and terminate the operation
prematurely.
Data Check-Bit 4

3. When command chaining is indicated, the


channel does not respond with subsequent
commands in a timely fashion.

4. A channel-discontinued retry is signalled on a


channel which has the 2860 Attachment
Feature installed, and the RPS command state
is O. See paragraph 3.12.

Overrun can occur when the total activity of the program


exceeds the capability of the channel. During output
operations, the Overrun condition indicates that data
recorded at the subsystem may be invalid. In these cases,
data transfer is stopped and the operation is terminated.

This bit indicates that one of the following errors has


been detected:

A correctable data error has been detected in


the information received from a disc drive (byte
2, bit 1 on; correction information is provided in
sense bytes 1 5 through 19).

3.7.3 Conditions Indicated by Bits


of Sense Bytes 1 through 23
For designations of bytes 1 through 23, and conditions
indicated by their individual bits, see paragraph 2.5.

3-17

@]

3.S READ DATA TRANSFER

1. After checking for CBR, the microprogram executes an RD = BI statement and loads a byte
into Buffer A. The Transfer Control Hardware
places this. byte - on the Bus In lines to the
channel and raises DATA IN. Because one
buffer now contains a byte of data, the
Q.EMPTY branch condition is reset in the
microprocessor.

3.S.1 Description
This paragraph provides a simplified block diagram
description of the Read Data Transfer operation.
Paragraph 3.8.2 provides a flowchart of this operation.
The alphabetical keys adjacent to lines of the block
diagram in Figure 3-7 are keyed to the text material
below.

DATA
FROM
DRIV E
INTER
FACE

I (J
JI
1

MICROPROCESSOR

I
~

.1

NUMBER OF BYTES TO BE TRANSFERRED

1. Before entering the Read Data Transfer sequence, the Channel Buffer Ready (CBR) and
the Queue-Empty (Q.EMPTY) branch conditions
in the microprogram are -set by Channel
Transfer bit 0 in the CHC Register being down
or reset (paragraph 6.2.2.5).

I
I
I

2. Microprogram loads the CC Register with the


number of bytes to be transferred (paragraph
6.2.2.5).
3. Loading the CC Register causes the CC
_to be reset in the microprocessor.

=0 latch

cc =

0
LATCH

I
I

MULTIPLEX
CHANNEL

1
1

1
(]0

a.EMPTY

(]~

RD = BI

(Q)

I
TRANSFER
CONTROL
HARDWARE

1. Microprogram initiates data transfers by setting


Channel Transfer bit 0 in the CHC Register. This
gates the two-byte buffer into the data transfer
path.

3. At this point in the control sequence, one of the


following occurs:

I
(J@
@

DATA
IN

a. Channel responds to DATA IN with DATA


OUT or COMMAND OUT. DATA OUT indicates that the byte has been accepted.
COMMAND OUT indicates truncation and
that the byte has been rejected by channel. A
response of DATA OUT causes the Q.EMPTY
branch condition to be set. If the CC Register
is decremented to zero, DATA IN and SERVICE IN are inhibited from being raised until
the CC Register is reloaded.

I SERVICE
IN
I
I

I
1~0

1
CC
REGISTER

!
I
I

!
I

2. Because there is still one available buffer for


more data, the CBR branch condition remains
set.

I
CHANNEL TRANSFER BIT AND
CHANNEL WRITE BIT

BUS
IN

~-

BUFFER
B (BII

CBR

@J01

lID

BUFFER
A (BII

CHC
REGISTER

L ______ J

b. Microprogram loads another byte into Buffer


B. This causes the CBR branch condition to
be reset in the microprocessor. Both buffers
then contain data that has nOlbeen accepted
by the channel. The microprogram cannot
load another byte of data into the buffers
until the channel accepts at least one of the
buffered bytes.

I
I

2. At the same time, the microprogram establishes


a read operation by resetting Channel Write bit
1 in the CHC Register.

[[]

[]

CHANNEL BUFFER READY AND


QUEUE EMPTY BRANCH CONDITIONS
When the CBR condition is set, it indicates that one
or both buffers have not been loaded by the
microprogram. When the Q.EMPTY condition is
set, it indicates that both buffers are empty.

MICROPROGRAM RESPONSE TO CBR

Figure 3-7. Read Data Transfer Block Diagram

DATA TRANSFER
Data is transferred to the channel by the Transfer
Control Hardware alternately raising DATA IN and
SERVICE IN to transfer successive bytes. Each time
a byte is transferred, the CC Register is
decremented to show how many bytes remain to
be transferred to the channel.

[J

END OF DATA TRANSFER


Transfer of data to channel is ended by one of the
following conditions:
1. When the CC Register is decremented to zero,
the CC = 0 latch is set in the microprocessor,
and the Transfer Control Hardware is inhibited
from raising SERVICE IN or DATA IN. This stops
transfer of data.
2. Channel truncates data transfer by raising
COMMAND OUT in response to SERVICE IN or

DATA IN. When this occurs, the CC Register is


not decremented, the a.EMPTY condition is not
set, and the CBR condition can be set or reset.
When the microprogram detects truncation, it
resets bit 0 in the CHC Register. This stops
transfer of data.

3. The channel issues a Halt I/O or Halt Device


command to the SCU by raising ADDRESS OUT.
This causes the microprogram to reset Channel
Transfer bit 0 in the CHC Register. This stops
the transfer of data.

@]

SENSE COMMANDS

For sense commands where data transfer is a read


operation over channel and data need not be
transferred regularly, the microprogram indicates
Suppressible Data Transfer by setting bit 5 in the
CHC Register. Transfer operation is identical to
transfer described in paragraphs ~
through
except that the Transfer Control
Hardware is inhibited from raising DATA IN or
SERVICE IN to transfer a new byte while SUPPRESS OUT is raised by the channel. When

[J ,

JPPRESS OUT is down, the Transfer Control


Hardware is not inhibited.

3.8.2 Flow Diagram


A flow diagram of the Read Data Transfer operation,
listing events that occur in the channel, channel interface, and microprogram, is shown in Figure 3-8. Each
event listed is preceded by a mnemonic that indicates
where the event occurs during the operation: channel
(C), channel interface (CI), or microprogram (M).

3-19

(CI)

BEFORE ENTERING SEQUENCE, CBR AND QEMPTY ARE BOTH


SET BY CHC REGISTER BIT 0 BEING DOWN OR RESET

~ (M)
1M)

Il PR O G
LOAD CC:
REG
(CI)

tIM)

(CI)

LOADING OF CC REGISTER RESETS CC


IN MICROPROGRAM

0 BRANCH CONDITION

(CI)

WHEN CC = 0, B(INT) BRANCH CONDITION IS SET IN


MICROPROGRAM TO NOTIFY MICROPROGRAM THAT ALL BYTES
HAVE BEEN TRANSFERRED

PRESENTED WHEN EITHER


DECREMENTED FROM 1 TO 0

CC

OR

CC

IS

BEING

MICROPROGRAM SETS CHC REGISTER BIT 0 ON AND RESETS


BIT 1
AFTER CHECKING FOR CBR. MICROPROGRAM EXECUTES RD = BI
STATEMENT WHICH LOADS DBUS INTO BI REGISTER.

HARDWARE DIRECTS DBUS INTO BUFFER B.

AFTER CHECKING FOR CBR. MICROPROGRAM EXECUTES RD


BI STATEMENT WHICH LOADS THE CONTENTS OF DBUS INTO
BI REGISTER

RAISE SERVICE IN
AND SET BYTE IN
BFR B ON BUS IN

INTERFACE CONTROLS SWITCHING BETWEEN BUFFERS A AND


B BUFFER A IS ALWAYS LOADED FIRST IN A SEQUENCE. Q.
EMPTY IS DEFINED AS BOTH BUFFERS EMPTY DURING A READ
OPERATION AND IS RESET WHEN EITHER BUFFER IS LOADED

~(M)

DECR CC
REG AND
SET
Q-EMPTY

1M)

BIT 0 IN CHC REGISTER IS CHANNEL TRANSFER BIT WHICH


INITIATES HARDWARE BUFFERING AND TAG CONTROL. BIT 1 IS
CHANNEL WRITE BIT THEREFORE, A READ OPERATION IS
DESIGNATED WHEN BIT 1 IS OFF

(M)

TO

HARDWARE DROPS DATA IN AND WAITS FOR MICROPROGRAM

TO LOAD BI REGISTER

(M)
(CI)

(2 OF 3)

MICROPROGRAM LOADS CC WITH TOTAL NUMBER OF BYTES


TO BE TRANSFERRED

CC REGISTER IS A SIXTEENBIT COUNTER WHICH CONTAINS THE


NUMBER OF BYTES TO BE TRANSFERRED

(M)
IlPROG
SET CHAN
XFER AND
CHAN READ

CBR AND QEMPTY BRANCH CONDITIONS ARE SET

......__- - - - - (C)

(CI)

GUARANTEES THAT DATA IS ON BUS IN BEFORE RAISING


SERVICE IN.

SERVICE OUT IS NORMAL RESPONSE BY CHANNEL TO SERVICE


IN AND SIGNIFIES DATA ACCEPTED

LOADING OF BUFFER A RESETS QEMPTY BRANCH CONDITION


IN MICROPROGRAM.

GUARANTEES THAT DATA IS ON BUS IN BEFORE RAISING DATA


IN

(C)

DATA OUT IS NORMAL RESPONSE BY CHANNEL TO DATA IN


AND SIGNIFIES DATA ACCEPTED.

(C)

LEADING EDGE OF DATA OUT CAUSES CC REGISTER TO BE


DECREMENTED AND SETS QEMPTY BRANCH CONDITION
(CI)

(M)

HARDWARE DROPS SERVICE IN AND


MICROPROGRAM TO LOAD BI REGISTER.

WAITS

FOR

aEMPTY BRANCH CONDITION IS SET IN MICROPROGRAM.

Figure 3-8. Read Data Transfer Flow Diagram (1 of 3)

3-20

FROM

(1 OF 3)

I
'RAISE DATA
IN AND
SET BYTE
IN BFR A
ON BUS IN

....I--+-- (el)

......Ir------ (el)

HARDWARE DROPS DATA IN AND SWITCHES MUX TO DIRECT


BYTE IN BUFFER B TO BUS IN

GUARANTEES THAT DATA IS ON BUS IN BEFORE RAISING DATA


IN

TO
(3 OF 3)

YES

>4"'_-1-__

(el)

CHECKING FOR NORMAL RESPONSE.

....Ir------ (el)

I'PROG
LOAD BFR
B AND RESET
CBR

DECR CC
REG AND
SET CBR

.....~--- (M)

AFTER CHECKING FOR CBR. MICROPROGRAM EXECUTES RD = BI


STATEMENT.

......1 - - - - -

HARDWARE GATES DBUS TO BUFFER B. BOTH BUFFERS ARE


NOW FULL. THEREFORE CHANNEL BUFFER READY (CBR) IS
RESET

....1 - - - - - - - (M)

MICROPROGRAM EXECUTES RD = BI.

....1 - - - - - ' - - -

(el)

HARDWARE GATES DBUS INTO EMPTY BUFFER AND RESETS


CBR

.>04------

(el)

CHECKING FOR NORMAL RESPONSE.

'----r----'
(el)

TO

HARDWARE WAITS FOR DATA TO GET TO BUS IN BEFORE


RAISING SERVICE IN .

(1 OF 31

">0.----

(el)

CHECKING FOR NORMAL RESPONSE. (QEMPTY NOT SET


BECAUSE BUFFER B IS STILL LOADED WITH DATA.)

(M)

CBR BRANCH CONDITION IS SET IN MICROPROGRAM.

......Ir------ (el)
TO

1----1.

HARDWARE DROPS SERVICE IN AND SWITCHES MUX TO DIRECT


BYTE IN BUFFER A TO BUS IN.

(3 OF 3)

Figure 38. Read Data Transfer Flow Diagram (2 of 3)

3-21

FROM
(1 OF 3)

~--+--- (e)

~-+---

(M)

INT REGISTER BIT 5 IS DEFINED AS TRUNCATION OR HALT 1/0


DURING DATA TRANSFER. IF BOTH BUFFERS ARE FULL AND THE
NORMAL RESPONSE IS NOT RECEIVED. MICROPROGRAM WILL
CHECK FOR THIS BIT BEING ON.

(~)

Q-EMPTY IS DEFINED AS BOTH BUFFERS EMPTY.

(M)

IF INT 5 IS NOT DETECTED, MICROPROGRAM MUST WAIT FOR


THE CHANNEL TO ACCEPT THE DATA IN THE BUFFERS BEFORE
PROCEEDING.

YES

YES

TURN OFF
CHNL XFER

~~------------------r-------------------

TURN OFF
CHNL XFER

(M)

MICROPROGRAM HAS FINISHED TRANSFER. MICROPROGRAM


NOW CHECKS FOR DATA VALIDITY MICROPROGRAM RESETS
CHC REGISTER BIT O.

(el)

RESET OF THIS BIT WILL PRESET Q-EMPTY, CBR, AND STOP


TRANSFER TAGS. IT ALSO RESETS HARDWARE SO THAT NEW
TRANSFER WILL BEGIN WITH DATA IN.

~~--------- (M)

MICROPROGRAM RESETS CHC REGISTER BIT 0 AND TERMINATES TRANSFER. HOWEVER, THE PROGRAM WILL CON
TINUE READING WITH NO TRANSFER TO THE CHANNEL TO
CHECK DATA VALIDITY.

~ .....- - - - - - - - - - - - - - - - - - - ; - - - - - - - - - - - - - - - - -

COMMAND
RETRY

DATA OVERRUN

I.~
__- - - - - -

(el)

HARDWARE RESPONDS EXACTLY AS ABOVE.

(M)

MICROPROGRAM CONTROLS ENDING STATUS PRESENTATION.

(el)

CHANNEL TRANSFER SUCCESSFULLY COMPLETED.

(M)

MICROPROGRAM CONTROLS COMMAND RETRY

(el)

CHANNEL TRANSFER ABNORMALLY TERMINATED.

Figure 3-8. Read Data Transfer Flow Diagram (3 of 3)

3-22

Control Hardware to raise DATA IN. Raising


DATA IN requests the first data byte fron"
channel.

3.9 WRITE DATA TRANSFER


3.9.1 Description

2. CBR remains reset until the first byte is


transferred from the channel.

This paragraph provides a simplified block diagram


description of the Write Data Transfer operation.
Paragraph 3.9.2 provides a flowchart of this operation.
The alphabetical keys adjacent to lines of the block
diagram in Figure 3-9 are keyed to the text material
below.

[[]

I
I

NUMBER OF BYTES TO BE TRANSFERRED


1. Microprogram loads the CC Register with
number of bytes to be transferred (paragraph
6.2.2.5).

BUS OUT

0 I
0

1. Microprogram initiates data transfers from the


channel to the channel interface by setting
Channel Transfer bit 0 in the CHC Register. This
gates the two-byte buffer into the data transfer
path.
2. At the same time, the microprogram establishes
a write operation by setting Channel Write bit 1
in the CHC Register.
CHANNEL BUFFER READY BRANCH CONDITION

10

TO A BUS

f
SERVICE IN

TRANSFER
CONTROL
HARDWARE

(]00

I
I

a.EMPTY

[)

RA BO

t t
cc

REGISTER

CHC
REGISTER

I
I

CBR

I
I

CC

0 LATCH

0@

I
I
I

0
@

I
I

L _ _ _ _ _ _ _ _ _ _ _ _ --.J

1. One instruction cycle after Channel Write bit 1


has been set. the Transfer Control Hardware
resets the CBR branch condition in the
microprocessor.
2 CBR reset condition indicates to the
microprogram that data is not yet available from
the channel.

TRANSFER OF FIRST DATA BYTE


1. When the channel transfers the first data byte,
the Transfer Control Hardware gates the byte
into one of the buffers.
2. Transfer Control Hardware then sets the CBR
branch condition to indicate to the
microprogram that data is available in one of the
buffers.

(Q)
@)~

I
I
I

'---

DATA IN

CHANNEL TRANSFER BIT AND CHANNEL WRITE


BIT

MULTIPLEX
CHANNEL

BUFFER
BIBO)

2. Loading the CC Register causes the CC = 0 latch


to be reset in the microprocessor.

rn:J

BUFFER
AIBO)

3. Because one buffer is still available for data,


Transfer Control Hardware immediately raises
SERVICE IN to request another data byte.

MICROPROCESSOR

...

[]

CONTROL SEQUENCE ALTERNATIVES


At this point in the control sequence, one of the
following occurs:
1. Microprogram responds to the CBR branch
condition by causing the Transfer Control
Hardware to gate the first data byte to Bus A.
This causes the CBR branch condition to be
reset.
2. Channel transfers another data byte to channel
interface (since one buffer remains empty). This
causes the CBR branch condition to remain set.
The Transfer Control Hardware cannot request
another byte from the channel until the
microprogram accepts at least one of the data
bytes. Each transfer of a data byte from the
channel causes the CC Register to be
decremented. If the CC Register is decremented
to zero, DATA IN and SERVICE IN are inhibited
from being raised until the CC Register is
reloaded.

REQUESTING FIRST DATA BYTE FROM CHANNEL


TRANSFER OF REMAINING DATA BYTES
1. Resetting the CBR condition. after setting
Channel Transfer bit 0 and Channel Write bit 1
with CC = 0 latch reset, causes the Transfer

Figure 3-9. Write Data Transfer Block Diagram

Data is transferred from the channel by the


Transfer Control Hardware alternately raising

3-23

DATA IN and SERVICE IN to transfer successive


bytes. Each time a byte is transferred, the CC
Register is decremented to show how many bytes
remain to be transferred from channel.

[8]

END OF DATA TRANSFER


Transfer of data from channel is ended by one of
the following conditions:
1. When the CC Register is decremented to zero
(data transfer completed). the microprogram
resets Channel Transfer bit 0 and Channel
Write bit 1 in the CHC Register. This stops
transfer of data.
2. Channel truncates data transfer by raising
COMMAND OUT in response to SERVICE IN or
DATA IN. When this occurs, the CC Register is

overruns, the Q.EMPTY branch condition in


microprocessor is set by the Transfer Control
Hardware when the present transfer sequence
has been completed and no new sequences are
to be initiated. When the Q.EMPTY branch condition IS set, the microprogram then resets
Channel Transfer bit 0 and Channel Write bit 1
in the CHC Register.

not decremented and the CBR condition is not


set (if it was reset). When the microprogram
detects truncation, it executes a Channel Freeze
control statement. This Channel Freeze control
statement essentially has no effect, but IS
executed because of timing requirements for
overrun conditions.
3. When the channel overruns the disc drive data
transfer rate by not responding to SERVICE IN or
DATA OUT before a data byte is required by the
SCU, the microprogram executes a Channel
Freeze control statement. Executing this control
statement causes the present SERVICE IN or
DATA IN transfer sequence to be completed,
and inhibits initiation of any new transfer
sequences. If SERVICE IN and DATA IN are both
reset when Channel Freeze is executed, the
data transfer is halted immediately. After
Channel Freeze control is executed on data

4. Channel issues a Halt I/O or Halt Device


command to the SCU by raising ADDRESS OUT.
This causes the microprogram to reset Channel
Transfer bit 0 and Channel Write bit 1 in the
CHC Register. This stops transfer of data.

[IJ

SUPPRESSIBLE DATA TRANSFER


For seek commands where data transfer is a write
operation over channel Gi'nd data need not be

transferred regularly, the microprogram indicates


Suppressible Data transfer to the Transfer Control
Hardware by setting bit 5 in the CHC Register. This
transfer operates identically to that described in
~ through (8], except that the Transfer Control
Hardware is inhibited from setting DATA IN or
SERVICE IN to transfer a new byte while SUPPRESS OUT is set at the channel. When SUPPRESS OUT is' down, the Transfer Control
Hardware is not inhibited.

3.9.2 Flow Diagram


A flow diagram of the Write Data Transfer operation,
listing events that occur in the channel. channel interface, and microprogram, is shown in Figure 3-10. Each
event is preceded by a mnemonic that indicates where
the event occurs during the operation: channel (C),
channel interface (CI), or microprogram (M).

3-24

FROM
(4 OF 4)

t'PROG
LOAD CC
REG AND
RSTCC=O
LATCH

RST CBR
BRCH
CONDITION

RT ~ CC DECODE LOADS T-BUS INTO CHANNEL COUNTER.

(CI)

CC REGISTER IS A TWO-BYTE COUNTER WITH A LATCHED


OUTPUT. THE LATCHED OUTPUT INDICATES 0 OR
DECREMENTED FROM 1 TO O. THE LATCHED INDICATOR IS
RESET WITH RT = CC DECODE.

(M)

RD = CHC DECODE CONNECTS D-BUS INTO CHANNEL CONTROL


REGISTER.

(CI)

CHC REGISTER BIT 0 IS THE CHANNEL TRANSFER BIT WHICH


ENABLES THE INTERFACE BUFFERS AND TRANSFER CONTROL
HARDWARE. CHC REGISTER BIT 1 IS THE WRITE BIT WHICH
TELLS THE HARDWARE TO ACCEPT DATA FROM THE CHANNEL.

(M)

CBR BRANCH CONDITION IS RESET IN MICROPROGRAM

(CI)

RAISE
DATA IN

(M)

t'PROG
SET CHNL
XFER AND
CHNL WRITE
BITS

FROM
(3 OF 4)

CBR BRANCH CONDITION INDICATES THAT EITHER ONE OR


BOTH BUFFERS ARE READY FOR MICROPROGRAM SERVICE. IN
A WRITE SEQUENCE. CBR IS RESET BY THE HARDWARE ONE
CYCLE AFTER CHC REGISTER BITS 0 AND 1 ARE SET ON.

(el)

DATA IN IS ALWAYS THE FIRST TAG RAISED BY THE HARDWARE


DURING TRANSFER

(C)

DATA OUT IS THE NORMAL CHANNEL RESPONSE AND INDICATES A BYTE OF DATA IS AVAILABLE ON BUS OUT

DROP
DATA IN

SET
BYTE
IN BFR A
INTO BO
REG

RAISE
SERVICE
IN

(CI)

AFTER LOADING DATA INTO BUFFER A. HARDWARE DROPS


DATA IN

(CI,

THE DATA IN BUFFER A IS SET INTO THE BUS OUT REGISTER SO


THAT IT IS SYNCHRONOUSLY AVAILABLE TO THE MICRO
PROCESSOR.

(el,

CC = 0 INDICATES THAT THE REQUESTED NUMBER OF BYTES


HAS BEEN ACCEPTED FROM THE CHANNEL.

(el,

AFTER CHECKING CC = 0 LATCH. HARDWARE RAISES SERVICE


IN.

(M)

MICROPROGRAM USES RA = BO STATEMENT TO TRANSFER


DATA FROM BO REGISTER TO DO REGISTER SO THAT IT CAN BE
SENT TO DRIVE CONTROLLER. CBR BRANCH CONDTION IS
RESET

(e)

SERVICE OUT IS THE NORMAL CHANNEL RESPONSE AND


INDICATES A BYTE OF DATA IS AVAILABLE ON BUS OUT.

(M)

MICROPROGRAM HAS A FIXED TIME IN WHICH TO RECEIVE


DATA FROM THE CHANNEL INTERFACE. THIS IS DETERMINED BY
THE FILE TRANSFER SPEED.

CBR BRANCH CONDITION IS SET IN MICROPROGRAM

NO

DECR
CC REG

SET CBR
AND LOAD
BYTE ON
BUS OUT
INTO BFR A

III

(M)

MICROPROGRAM TIMEOUT REFERS TO THE TIME WHEN THE


DRIVE CONTROLLER MUST HAVE DATA. OR AN OVERRUN WILL
OCCUR

(M)

CBR BRANCH CONDITION IS SET IN MICROPROGRAM

(M)

(CI)

AFTER THE CHANNEL RAISES DATA OUT. THE HARDWARE


LOADS BUFFER A WITH THE DATA ON BUS OUT AND SETS THE
CBR BRANCH CONDITION

(el,

SETTING OF CBR INDICATES THAT DATA IS AVAILABLE IN ONE


OF THE BUFFERS. DATA ON BUS OUT IS LOADED INTO BUFFER
B.

Figure 3-10. Write Data Transfer Flow Diagram (1 of 4)

3-25

FROM
(1 OF 4)

FROM
(3 OF 4)

DROP
SERVICE
IN

SET BYTE
IN BFR B
INTO BO
REG

....1 - - - (el)

AFTER LOADING DATA INTO BUFFER B. HARDWARE DROPS


SERVICE IN

, ........- - - leI)

HARDWARE SETS BYTE IN BUFFER B INTO BO REGISTER SO


THAT IT IS SYNCHRONOUSLY AVAILABLE TO THE
MICROPROCESSOR.

~------- (e)

DATA OUT IS NORMAL CHANNEL RESPONSE AND INDICATES A


BYTE OF DATA IS AVAILABLE ON BUS OUT.

' - - - - - (M)

MICROPROGRAM HAS A FIXED TIME IN WHICH TO RECEIVE


DATA FROM THE CHANNEL INTERFACE: THIS IS DETERMINED BY
THE FILE TRANSFER SPEED.

.....1 - - - - + - - - - - - - - (el)

SETIING OF CBR INDICATES THAT DATA IS AVAILABLE IN ONE


OF THE BUFFERS DATA ON BUS OUT IS LOADED INTO BUFFER
A.

1------+-------1..-

CBR BRANCH CONDITION IS SET IN MICROPROGRAM.

YES
DECR
CC REG

YES

'--_ _ _ _ lei)

CC = 0 INDICATES THAT THE REQUESTED NUMBER OF BYTES


HAS BEEN ACCEPTED FROM THE CHANNEL.

NO

SET CBR
AND LOAD
BYTE ON
BUS OUT
INTO BFR A

RAISE
DATA
IN

NO

=0

....F - - - - (el)

AFTER CHECKING CC

, - - - - - (M)

MICROPROGRAM USES RA = BO STATEMENT TO TRANSFER


DATA FROM BO REGISTER TO DO REGISTER SO THAT IT CAN BE
SENT TO DRIVE CONTROLLER. CBR BRANCH CONDITION IS
RESET.

(M)

LATCH. HARDWARE RAISES DATA IN.

TO
(3 OF 4)

TO

TO

(4 OF 4)

(4 OF 4)

Figure 3-10. Write Data Transfer Flow Diagram (2 of 4)

3-26

FROM
12 OF 4,

FROM
., OF 4

FROM
14 OF 41

r - - _......._-""
DROP
DATA'
IN

SET BYTE
IN BFR A
INTO BO
REG

RAISE
SERVICE
IN

(CII

AFTER LOADING BUFFER A. HARDWARE DROPS DATA IN

(CI)

HARDWARE SETS BUFFER A BYTE INTO BO REGISTER.

(CI)

CC
0 INDICATES THAT REQUESTED NUMBER OF BYTES HAS
BEEN ACCEPTED FROM THE CHANNEL

(ell

(CI)

DATA ON BUS OUT IS LOADED INTO BUFFER B.

DROP
SERVICE
IN

(CI)

AFTER LOADING DATA INTO BUFFER B. HARDWARE DROPS


SERVICE IN.

DECR
CC REG

(CI)

HARDWARE LOADS BYTE ON BUS OUT INTO BUFFER B. BOTH


BUFFERS ARE NOW FULL. THEREFORE. HARDWARE WILL NOT
RAISE ANY TRANSFER TAGS UNTIL ONE BUFFER IS EMPTY.

(CI)

CC = 0 INDICATES THAT REQUESTED NUMBER OF BYTES HAS


BEEN ACCEPTED FROM THE CHANNEL.

HARDWARE RAISES SERVICE IN.

(M)

(C)

LOAD
BYTE ON BUS
OUT INTO
BFR B

MICROPROGRAM USES RA
BO STATEMENT TO TRANSFER
DATA FROM BO REGISTER TO DO REGISTER SO THAT IT CAN BE
SENT TO DRIVE CONTROLLER.CBR BRANCH CONDITION IS
RESET.

(M)

IN THIS CASE. BOTH BUFFERS A AND B ARE LOADED WrrH


DATA. THE MICROPROGRAM MUST SERVICE THE BO REGISTER
TO ALLOW DATA TRANSER TO CONTINUE.

WHEN CHANNEL RAISES SERVICE OUT. IT INDICATES THAT A


BYTE OF DATA IS AVAILABLE ON BUS OUT.

Figure 3-10. Write Data Transfer Flow Diagram (3 of 4)

3-27

FROM

FROM
C (1 OF
G (2 OF
K (3 OF
N (3 OF
Q (4 OF

B (1 OF 4)
F (1 OF 4)
I (2 OF 41

'----+---- (M)

MICROPROGRAM USES RA
BO STATEMENT TO TRANSFER
DATA FROM BO REGISTER TO DO REGISTER SO THAT IT CAN BE
SENT TO DRIVE CONTROLLER CBR BRANCH CONDITION IS
RESET

' - - - - - - - - (e)

DATA OUT IS THE NORMAL CHANNEL RESPONSE AND INDICATES A BYTE OF DATA IS AVAILABLE ON BUS OUT

LOAD BYTE
ON BUS OUT
INTO BFR A

.........- - - - - (el)

DATA LOADED ON BUS OUT IS LOADED INTO BUFFER A.

DROP
DATA
IN

....~----- (el)

AFTER LOADING DATA INTO BUFFER A


DATA IN

HARDWARE DROPS

/lPROG
SET CHNL
FREEZE

4)
4)
4)
4)
41

INHIBIT
DATA IN
AND
SERVICE IN

.~.....- - - - (el)

NORMAL SHUTDOWN OF DATA TRANSFER. CC ~ 0 WILL INHIBIT


THE RAISING OF EITHER TAG BY HARDWARE

/lPROG
UNLOADS
DATA FROM
BUFFERS

.....1 - - - -

(M)

MICROPROGRAM TAKES DATA FROM BO REGISTER UNTIL CBR


IS RESET

/lPROG
TURN OFF
CHNL XFER
AND CHNL
WRITE

....~--- (M)

MICROPROGRAM HALTS ALL FURTHER TRANSFER BY RESETTING CHANNEL TRANSFER BIT AND CHANNEL WRITE BIT IN CHC
REGISTER

.....t - - - -

MICROPROGRAM CHECKS FOR


PROCEEDING TO ENDING STATUS

(M)

'~"'''---------------(M)

........~-------------+----(M)

~------

(ell

BEFORE

CHANNEL FREEZE DOES NOT ALLOW THE INTERFACE TO RAISE


ANY NEW TAGS_ BUT ALLOWS ANY TRANSFER IN PROGRESS TO
BE COMPLETED

CH INT (INT REGISTER BIT 51 INDICATES EITHER TRUNCATION OR


HALT I/O. IN EITHER CASE, TRANSFER OF DATA FROM THE
CHANNEL WILL BE STOPPED

CC = 0 INDICATES THAT REQUESTED NUMBER OF BYTES HAS


BEEN ACCEPTED FROM THE CHANNEL

~------- (M)

BOTH BUFFERS ARE FULL. DATA MUST


MICROPROGRAM TO CONTINUE TRANSFER.

BE

TAKEN

BY
ERROR
RECOVERY

MICROPROGRAM WAITS FOR Q-EMPTY BEFORE TURNING OFF


CHANNEL TRANSFER. THIS GUARANTEES THAT ALL TRANSFER
TAGS ARE DOWN.

(el)
' - - - - - - - - (M)

VALIDITY

RF 7 CH FREEZE STATEMENT TELLS THE HARDWARE TO ACCEPT


THE BYTE NOW BEING TRANSFERRED AND THEN INHIBIT
FURTHER TRANSFER

(el)

DECR
CC REG

DATA

DURING A WRITE
FOLLOWING:

OPERATION,

Q-EMPTY

INDICATES THE

1. THE CHANNEL IS QUIET AND ALL CHANNEL TRANSFER TAGS


ARE DOWN.
2. CC =0 AND ALL CHANNEL INTERFACE TRANSFER TAGS ARE
DOWN.

Figure 3-10. Write Data Transfer Flow Diagram (4 of 4)

3-28

3.10 SCU-INITIATED CHECK 1 ERROR


CONTROL SEQUENCE (POLLING),
SIMPLIFIED

OPERATIONAL IN UP

The flowchart and simplified logic diagram shown in


Figure 3-11 illustrate an SCU-Initiated Check 1 Error
Control Sequence. The flowchart and logic diagram are
keyed to the text material below.

COMMAND OUT UP
ADDRESS IN UP

=------1==;:====1

DISCONNECT IN UP

INHIBIT
DISCONNECT
LATCH

SELECT RESET UP
START

00~
OPERATIONAL IN UP

POWER ON RESET
ANY
SCU
OPERATIONAL
IN UP

NOT ANY
OPERATIONAL IN UP

YES

CHECK 1 UP

~Jo

SCU RAISES
REQUEST IN
TO r:HNL

CHNL RAISES
CMD OUT

SCU RAISES
DISCONNECT
IN AND DROPS
ADDR IN

SCU PERFORMS
SELECTIVE RST
AND DROPS
DISCONNECT IN
AND OPERATIONAL
IN

CHNL RAISES
SEL OUT AND
HOLD OUT

SCU RAISES
OPERATIONAL
IN AND DROPS
REQUEST IN

CHNL INITIATES
A SELECTIVE
RST

I
SCU RAISES
ADDR IN
AND PLACES
ADDR ON
BUS IN

[]

SCU RAISES
DISCONNECT IN

CHNL INITIATES
A SELECTIVE
RST

SCU has a Check 1 error. If no channel in this SCU


has its OPERATIONAL IN up, REQUEST IN is raised
to the channel.

@]

S(,_ PERFORMS
SELECTIVE
RST AND DROPS
DISCONNECT
IN AND
OPERATIONAL IN

Channel responds to REQUEST IN with SELECT


OUT and HOLD OUT
SCU raises OPERATIONAL IN and drops REQUEST
IN.
SCU raises ADDRESS IN, and places address of
SCU and a 000 Device Address on the Bus In lines.

SELECT
OUT LATCH SET
SCAN A UP
ADDRESS IN UP

MACHINE RESET
COMMAND OUT UP

Channel stores address and responds with COMMAND OUT.


SCU raises DISCONNECT IN and drops ADDRESS
IN

If the SCU with the Check 1 error is already


selected (its OPERATIONAL IN is up) and its address
has already been given to the channel, the control
sequence proceeds as follows:
SCU raises DISCONNECT IN.

Channel initiates a selective reset by raising SUPPRESS OUT and dropping OPERATIONAL OUT.

Channel initiates a selective reset by raising SUPPRESS OUT and dropping OPERATIONAL OUT.

SCU performs selective reset and drops DISCONNECT IN and OPERATIONAL IN.

SCU performs a selective reset and


DISCONNECT IN and OPERATIONAL IN

drops

Figure 3-11. Check 1 Error Controlling Sequence Flowchart and Logic Diagram

3-29

3.11 MULTI-CHANNEL SWITCH


OPERATION
The Multi-Channel Switch (MCS) feature permits the
SCU to be shared by up to four selector or block
mUltiplexer channels in any combination; i.e., four selector channels, four block multiplexer channels, or any
combination thereof. The channels may be attached to
the same or different central processing units. Anyone
of the channels may be reserved exclusively by any of the
drives attached to the SCU. Channel switching and
device reservation are controlled by the channel
program. Two special commands are invoked for MCS
operation: Device Reserve and Device Release. Refer to
the MCS logic shown in Figure 3~ 12 throughout the
discussion that follows.

Chaining is indicated without Device End in the


status byte, and the channel does not disconnect.

Chaining is indicated and Device End is included


in the status byte.

Ending status associated with an interface disconnect has not been accepted by the channel.

When a channel connection is maintained as a result of


a contingent connection, the SCU will not respond to
polling by the channel, except to present Control Unit
End. When the contingent connection is terminated, the
SCU can be selected by any channel.
The length of time the SCU is connected to a channel
determines how the channel selection switch responds
to a channel-initiated sequence from the other channels.
The channel/SCU connection is classified as "instantaneous" or "long."

3.11.1 Channel Selection Switch

NOTE

Once the SCU has been selected by any channel, it is


reserved by that channel until Ending Status is
presented. The SCU can then be selected by any channel
unless:

The last status byte was part of a channelInitiated signal sequence and was stacked by
the channel.

A contingent connection is established.

Chaining is indicated and a format write operation is in progress.

Chaining is indicated without Device End in the


status byte, the channel disconnects, and the
SCU becomes busy to allow:

In the following descriptions, the


conditions of any of the channels
may be interchanged.

3.11.2 Instantaneous Connection


At the moment the channel selection switch connects
the SCU to a channel, the connection is considered to be
instantaneous until otherwise determined by the SCU.
The state of a channel connection changes from instantaneous to long if:

Channel is processing a Start I/O instruction, or

Channel indicates chaining on Device End during a polling sequence.

a. Completion of a format write operation.


b. Execution of a Diagnostic Load or Diagnostic
Write command.

If a channel/SCU connection is caused by a condition


other than the above, the state of the connection remains
instantaneous until the operation is complete.

c. Execution
procedure.

During an instantaneous connection on channel A, the


selection logic does not respond to a channel-initiated

of

an

SCU

error

recovery

sequence on any other channel until the instantaneous


connection is terminated. When the instantaneous connection terminates, the selection logic either remains
connected to channel A (if the connection state changed
from instantaneous to long), or the SCU becomes
available to all channels (neutral). If the switch returns to
neutral from channel A during a channel-initiated selection sequence on any other channel, the logic immediately connects to the applicable channel. If the
connection state changed from instantaneous to long on
channel A, the response to the selection logic is as
defined in paragraph 3.11.3.

3.11.3 Long Connection


During a long connection on channel A, the SCU
responds with a short Control Unit Busy Sequence (busy
and status modifier) to a channel-initiated selection
sequence on any other channel. Whenever the short
Control Unit Busy Sequence occurs, the microprogram
attempts to present Control Unit End status to the
appropriate channel once the selection logic returns to
neutral. The address byte associated with this status is
the address of the lowest numerically addressed device
which is not implicitly connected to another interface.
The pending Control Unit End status does not cause the
SCU to appear busy to channel A as long as the selection
switch is not actually connected to anyone of the other
channels.

When the switch is in the UNTAG position, Device End


for the not-ready-to-ready sequence is also presented to
all channels. However, this Device End is cleared as soon
as it is accepted by any channel, and no further attempt
is made to present Device End to any other channel.
When a device is busy for any reason (including reservation to channel A), any command from another channel
addressed to that device will be rejected with a Busy
status. This, in turn, causes the SCU to attempt sending
a status byte containing Device End to the appropriate
channel after the busy condition has been terminated.
The address byte associated with this status byte will be
the same as that associated with the busy status byte.
Device End status resulting from any channel command
will be presented to the channel that issued the command.

3.11.5 Addressing
The base address (four high-order bits for 16-spindle
configuration, three for 32-spindle configuration) of the
storage control on one channel is independent of the
base addresses on the other channels. However, the
three low-order address bits for any attached device
must be the same on all channels.

3.11.4 Device Status

3.11.6 Resets

Device End status resulting from a not-ready-to-ready


transition will be presented under control of the
TAG/UNTAG switch. A not-ready-to-ready sequence
Device End occurs after a disc pack change is completed
and the affected module is returned to an on-line
condition.

System Reset clears all reservations and status conditions stored in the SCU for the resetting channel,
terminates all block mUltiplex command chains in
progress on the resetting channel, and resets all device
interrupts not associated with the other channels. The
reset may be initiated by any channel at any time.
Reservations, status, and device interrupts for the other
channels, as well as block multiplex chains in progress
on the other channels, are not affected. If a channel
initiates a System Reset while the selection logic is
connected to another channel, a Machine Reset is
performed when the selection logic goes to neutral. A
Selective Reset has no effect on device reservations or
status.

When the switch is in the TAG position, Device End for


the not-ready-to-ready sequence is presented to all
channels. After Device End is accepted by a particular
channel, the module can be addressed for command
execution by that channel. Before any other channel can
use the module, it must also receive the not-ready-toready sequence Device End.

3-30

11\1111\1 Sill l'll~)N ~\

SELECT A
l;[

NlllAl HESE T

ALL LOGIC FOR THESE TWO CIRCUITS CONTAINED ON CIS02.

I'HIOHITY WINDOW A
ICP REGISTER BIT 7)
CASOI
FFINOTSET = D INIT IAl Sf LECTION B

SELECT B

SELECT C
SELECT A

FF2 NOT SET

SELECTC
SELECT D
FFI

FF2 NOT SET

FF1SET=D-

SCAN A

SCAN B
SCANS A THROUGH 0 CONTROL GENERA
TION OF SHORT SCU BUSY SEQUENCE
AND PROPAGATION OF SELECT OUT.

GENERAL RESET
PRIORITY WINDOW B
(CP REGISTER BIT 6)

FF1NOTSET = D -

SELECT A
SELECT B
FFI

CBSOI

FF2

SCAN C

FF2 SET
TO CI REGISTER
BIT6ANO
SCAN LOGIC

INITIAL SELECTION C

FF1SET=D-

SCAN 0

FF2 SET
SELECT C

SELECT D
SELECT C
FFI

GENERAL RESET
SELECT B
SELECT A
FFI

PRIORITY WINDOW C
(CP REGISTER BIT 5)
CCSOI

CI REGISTER - BIT6

SELECT B
SELECT 0

INITIAL SELECTION 0

r-_
106

SELECT 0

FF1
SELECT A
SELECT B

GENERAL RESET

PRIORITY WINDOW D
(CP REGISTER B(T 4)

CHC REGISTER
BIT6
TO CI REGISTER
BIT 7 AND
SCAN LOGIC

COS01
CI REGISTER
BIT 7

SELECT C
SELECT 0

_-,--_r-_

106 AND 1D7 CONTROL MULTIPLEXING


OF CHANNEL BRANCH CONTROLS AND
CHANNEL DATA TO AND FROM
MICROPROCESSOR.
107

ANY OPERATIONAL IN UP
LOCK CHANNEL SWITCH
(CHC REGISTER BIT 4 SET)

CHC REGISTER --'--~-


BIT7

It, OSCILLATOR CLOCK

ANY OPERATIONAL IN UP
ANY SCU SHORT BUSY
SEQUENCE IN PROGRESS
SCU BUSY
(CHC REGISTER BIT 3 SET)

LOCK CHANNEL SWITCH


(CHC REGISTE R BIT 4 SET)

Figure 3-12. Multi-Channel Switching Logic

3-31

3.12 2860 SELECTOR CHANNEL


ATTACHMENT FEATURE

3.1 2.1 Description


The 3672 2860 Attachment Feature provides the
capability for the Memorex 3672 Storage Control Unit
(SCU) to operate in a simulated block multiplex mode
when connected to an IBM 2860 Selector Channel. The
objective of simulating block multiplex operations requires that channel command chaining be broken to
effect a channel disconnection (to provide channel and
SCU availability) .and that, upon the SCU signalling the
channel with Retry status,' the last command be
retransmitted to the SCU to provide error recovery
procedures. On actual block multiplex channels,
Disconnected Command Chaining (DCC) and Command
Retry (CR) are performed by interaction between the SCU
and the channel without CPU interaction. However, the
2860 Selector Channel does not support DCC or CR, and
if a SCU without the 3672 2860 Feature attempts either
of these sequences on a 2860 channel the following
occur, respective'ly:
a. The 2860 maintains connection to the SCU
upon receipt of Channel End (CE) only status.
b. The 2860 inhibits chaining upon receipt of Retry
status, that is, Status Mcidifier (SM), Channel
End (CE), and Unit Check (UC). This causes the
SCU to discontinue the retry sequence and
indicates Equipment Check in a subsequent
Sense command.
In order to simulate block multiplex operation, the
software in the using System/360 must provide the
necessary interaction with the SCU to perform DCC
and CR as opposed to the channel/CU interaction that
occurs on actual block multiplex channels. As an
example, typical channel programs for the 3670/5 disc
on block multiplex channels (IBM-OS release 21.6) are
of the following form:

3672 .21 -0001-10175

Seek
Set File Mask
Set Sector
Search
Write
Set Sector
Search
Read

To support the SCU on the 2860 Selector Channel, the


using software system must include a new command,
Set RPS, in every channel program for the 3670/5. This
command causes the SCU to enter the "RPS Mode" and
unique status is presented to force a channel disconnection or command retry sequence on the 2860 channel. A
channel program for the 3670/5 then takes on the
following form:
Set RPS
Seek
Set File Mask
Set Sector
Search
Write
Set Sector
Search
Read

The Start I/O instruction is executed by the CPU and


receipt of Condition Code 0 in the Program Status Word
(PSW) indicates that the channel is executing the
channel program. The SCU executes the channel
commands beginning with the Set RPS command.
Upon receipt of the Seek command, the SCU presents
disconnect status to the channel if mechanical motion
is required. The channel disconnects and the SCU
"marks" the seeking device as being in the DCC state,
even though the channel program was broken when the
channel did not indicate command chaining when it
accepted the disconnect status. The SCU is now
available for operations with other connected devices
which are not busy or offline.

The SCU tests for the seek complete condition at the


devices when it is in the idle state and initiates a
selection to the channel when the seek complete condition is detected. The channel selects the SCU and stacks
the seek complete status, Device End (DE). The SCU is
available after DE is stacked for Start I/O, Test I/O, or
Halt I/O on other devices. The SCU signals a request for
reconnection with DE status, and the channel interrupts
the CPU and generates a Test I/O to.obtain device status
from the SCU. DE status is placed in the Channel Status
Word (CSW) and the CSW is stored in main storage
during the I/O interrupt sequence.
When DE ending status to the Test I/O instruction is
accepted,the channel does not indicate chaining. Instead,
the SCU maintains a contingent connection and waits for
the restart command. A selection of another device will
not be accepted and UC will be presented in initial status.
The SCU reconnection on an actual block multiplex
channel is similar to the above except that the channel
indicates chaining to DE status and the chained
. reselection loop is entered to wait for the new command
with the Suppress Out tag (chaining indication) active
until the channel reselects the SCU for the next
command. The Suppress Out tag will not be active on
the 2860 selector channel because the channel processing an interrupt to the CPU rather than a channel
hardware reconnection.
To effect reconnection, the using software system continues the original command chain with the Set RPS
command prefixed to it. The Start I/O instruction required to continue the command chain then points to the
following command sequence:
Set RPS
Set File Mask
Set Sector
Search
Write
Set Sector
Search
Read

The receipt of Condition Code 0 indicates that the


channel program has been restarted. The channel
program continues until the next disconnect occurs. For
the
example
being
analyzed,
tvito
more
disconnect/reconnect sequences would take place (in
the Set Sector commands) identical to that described
above for the Seek command.
Up to four channels in any combination of block multiplexer and 2860 may be attached to the SCU. Support
of the Reserve and Release commands associated with
the SCU Multichannel Switch Feature is provided for the
single channel and multichannel versions of the SCU
when the 367212860 Attachment Feature is present.
When channel programs are not prefixed with the new
Set RPS command on a 2860 channel, the SCU does not
break the channel program for CE only disconnection and
some types of CR disconnection; instead, channel
programs are executed to the end of the chain. No seek
or RPS overlap is available in this mode and the CR
procedure for error recovery is not attempted.

3.12.2 Functional Characteristics


3.12.2.1 GENERAL
The functional characteristics of the SCU with the 2860
Attachment Feature are identical to the basic SCU
characteristics. except as follows:
1. The SCU' will accept and execute the new Set
RPS command on all 2860 interfaces.
2. If the $et RPS command has been issued to the
SCU, disconnect status of CE, UC, and UE is
presented on 2860 interfaces instead of CE
only. In addition, the SCU will maintain DCC and
CR on 2860 interfaces even though chaining is
not indicated by the channel(s).
The following sections describe
differences for 2860 interfaces only.

in

detail

these

3-32

3.12.2.2 CHANNEL COMMANDS


CONTROL COMMANDS
SET RPS
Command Code 2F (hex)

Figure 3-13

Changes the state of the SCU to RPS mode, and


coordinates the simulated Disconnected Command
Chaining and Command Retry between the SCU and
the CPU
The Set RPS command is valid only on 2860 Selector
channels attached to the SCU. If this command is
received from a block multiplex channel attached to the
SCU, Unit Check will be presented in initial status and
Command Reject/Invalid Command will be indicated in
a subsequent Sense command. If the Set RPS
command is truncated, the SCU will give Unit Check
and the subsequent sense will indicate Command
Reject/CCW Count Less Than Required.
Execution of the Set RPS command causes a state code
byte to be transferred to the SCU. Bits 6 and 7 of the
state code define the operation to be performed by the
command. Bits 0-5 must be zero.

10

Reconnect after disconnection


(CE only)

11

Reconnect after command retry


disconnection

Upon receipt of the Set RPS command, the SCU checks


the state of the 3670/75 and the state code received in
the one-byte argument. When the state code does not
correspond to the 3670175 state, the SCU posts ending
status of (CE, DE, UC) and a subsequent Sense will
indicate Command Reject/Invalid Sequence.

3670 State

IS

SET RPS

NO

-=

Definition
No such state

OBTAIN RPS
STATE BYTE
FROM
CHANNEL

Initialize RPS mode


2

Disconnected command chaining

Command retry disconnected

When the 3670/75 is.in state 2 or 3, the receipt of any


command, except Set RPS or Test liD, will cause UC in
initial status and Command Reject/Invalid Sequence in
a subsequent Sense.

Bits 6. 7

Operation

00

Not valid. ending status of CEo


DE. UC (command reject). data
value not as required

DATA ADDRESS-Specifies main storage location of the


state code byte

Set RPS mode

COUNT-One

01

THE SET RPS command


IS valid only If the 2860
Attachment Feature
installed.

NO

FLAGS-Used at discretion of the programmer

304

SET
INVALID
COMMAND

036

SET
STATE 1

SET
CMD REJECT.
INVALID DATA

078

Figure 3-13. Set RPS Flow Diagram

3672.21 -0001-10175

3-33

Seek, Seek Cylinder, Recalibrate

Sense

These commands present disconnect ending status


(CE, UC, UE) when access motion is required. The
command operation is unchanged but the channel
sequence is modified when the SCU is in RPS mode.
The completion of the access motion results in an SCUinitiated selection and a subsequent CPU interrupt
when DE is posted. The DE interrupt is the signal that
the device is no longer busy and the channel program
may be restarted. Reconnection of the channel program
requires prefixing a Set RPS command with a state
code byte of state 2 to the channel program. The restart
address in the channel program is specified by the
CCW address in the CSW which was stored when the
disconnect interrupt occurred.

All sense command operations are unchanged.


Read
All read command operations are unchanged.

Reconnect Sequence

Write

Operations that occur during a reconnect sequence are


as follows:

The Set Sector command presents disconnect ending


status (CE, UC, UE) when rotational positioning is
required. Rotational positioning is not required when
the current disc's rotational position is equal to the
target sector minus seven; in this case, the ending
status is CE and DE
The SCU will raise the Request In tag to initiate
reconnection when the SCU is in the wait loop and the
drive interrupt is active. The drive interrupt is active for
one sector period (130 /-Lsec) on each disc revolution.
The sector corresponding to the drive interrupt period
is the sector specified by the Set Sector command
minus 7. The reconnect request on the 2860 is therefore
5 sectors earlier than the corresponding request for a
block multiplex channel. This allows for an additional
delay to accomplish the software analYSis of the Device
End status (I/O interrupt) and execution of the
reconnect command Set RPS.
The SCU will raise the channel Request In tag to initiate
reconnection when anyone of the attached disc drives'
rotational position corresponds to the Set Sector
argument minus 7.

Search
All search command operations are unchanged.

The 2860 Attachment Feature allows the software in the


CPU to simulate disconnected command chaining (DCC).
The SCU operation is modified when in RPS mode as
follows.
Disconnect Status
Disconnect status is changed from CE only to (CE, UC,
UE). (The 2860 maintains connection to the SCU when
CE only status is presented.) When the new status is
presented, the 2860 disconnects from the SCU and
posts the status to the CPU and a I/O interrupt.

Chaining While Disconnected


The SCU receives the not chaining indication from the
2860 when disconnect status of (CE, UC, UE) is
presented, but the "disconnected and chaining"
housekeeping is performed to simulate the disconnected command chaining (DCC) state. This is
done to maintain the current drive status data so that
the command chaining may be restarted when the
mechanical motion has completed. When reconnection
IS initiated and the Start I/O is issued, the SCU will not
treat the command as the start of a new chain, but will
treat it as if it were chained to the previous command
(which caused the disconnect).

Diagnostic Load, Diagnostic Write

a. SCU completes a channel sequence and enters


the wait loop (idle state).
b. SCU detects a drive interrupt.
c. SCU raises Request In.
d. 2860 polls (sequences channel tags by raising
SELECT OUT while ADDRESS OUT is down).

~n7221

-0001-10175

p. 2860 receives Start 110 from CPU for the


reconnect channel program on the device which
presented the reconnect interrupt.

The DE status posted by the SCU (connected to a 2860)


to reconnect to the channel is unchanged. The 2860

q. SCU processes the Set RPS command whose


state code byte indicates state 2 '(reconnect) and
presents ending status of CE, DE.
r. 2860 accepts status and indicates chaining.

s. 2860 continues channel program.


t. SCU continues channel program.

e. SCU detects poll and performs selection sequence with channel using the device address
of the lowest logical numbered drive in the DCC
state.

3.12.2.4 Error Recovery

f. SCU presents DE status to channel and channel


indicates stack status to SCU.

Error recovery procedures are changed to simulate


command retry disconnect and reconnect sequences
for 2860 channels only.

g. SCU disconnects from the channel and maintains the DCC state of the drive. The stacked
status state is not entered because the "Device
End pending" from the drive interrupt is unchanged.
h. SCU enters the wait loop and raises Request In
when a drive interrupt is present at one or more
of the drives.
j.

2860 attempts to obtain an 1/0 interrupt for the


drive whose status was stacked.

Command Retry Status


The command retry status of (SM, CE, UC) is unchanged. The 2860 does not support command retry and
breaks the command chain when this status is
presented by the SCU. The 2860 posts this status to the
CPU with a I/O interrupt.

Chaining While Disconnected For Command Retry

k. 2860 obtains interrupt from CPU.


2860 performs Test 110 (channel hardware
generated) on device whose status was
previously stacked and completes the CPU interrupt sequence.

Reconnect Status
The operation of these commands is not affected by the
RPS mode of the SCU. They do not present disconnect
status.

n. SCU enters contingent connection wait loop. All


Request In tags are down so no poll selection
will occur. Channel-initiated selection of any
other device will be rejected with UC status and
any selection from the other channels will get a
control unit busy.

All write command operations are unchanged.

3.12.2.3 DISCONNECTED COMMAND CHAINING


Set Sector

does not Indicate chained reconnect ion by indicating


chaining when status is presented and the SCU must
maintain a contingent connection on the device until the
CPU software provides the simulated reconnection.

m. SCU posts DE status and resets interrupt at


drive. Status is accepted but chaining is not
indicated by channel.

The SCU receives the not chaining indication from the


2860 when the command retry disconnect status of
(SM, CE, UC) is presented, but error recovery processing is performed with the chaining state simulated.
When the reconnection is initiated and the Set RPS
command is processed, the command chained to the
Set RPS command will be treated as the retry command
in the same manner as would occur for normal retry
command processing on a block multiplex channel.

3-34

Cllllllll.lnd Retry Reconnect Status

Disconnect Status

lilt' status of DE posted by the SCU to reconnect to the


channel IS unchanged. Although the 2860 does not
Indicate chained reconnection by indicating chaining
when status is presented, the SCU maintains the
pseudo chaining state of the device until the CPU
software provides the simulated reconnection.

The SCU posts CE-only status for disconnect, but the


2860 does not disconnect. The result is that the channel
remains connected to the SCU until mechanical motion
is completed. No RPS or seek overlap is available except
for the case where chaining is not indicated when CE
status is posted. In this case, the 2860 presents CE to the
CPU with an I/O interrupt. After the motion has completed, the SCU presents DE to the channel and another
110 interrupt is taken.

Reconnect Sequence
The reconnect sequence for a command retry disconnect
IS similar to that of a normal disconnect. After the Set
RPS command with state code byte indicating state 3 is
processed, the error recovery sequence is continued as
below:
a. DE status is accepted by the 2860
b. SCU enters a pseudo reconnect wait loop. This
is equivalent to the UC contingent connection
wait loop.
c. SCU receives the Set RPS command for command retry reconnect and presents CE, DE
ending status.
d. SCU receives chaining indication from the 2860
and enters the chained reselection loop.
e. SCU receives the retry command and enters the
command retry procedure to perform the retry of
the command.
f. SCU continues the channel program.
g. 2860 continues the channel program.

3.12.2.5 NON-RPS MODE OPERATION


Block mUltiplex channels always operate in the "not RPS
mode" and there is no effect upon these channels when
the 2860 Attachment Feature is present. The attached
2860 channels are presented with different status for
error recovery operations in the SCU to provide compatibility with software which cannot accept the command retry status on an I/O interrupt. The SCU does not
attempt to provide disconnected command chaining for
overlap of mechanical motion or for error recovery
operations.

3672.21-0001-10/75

Command Retry Status


The SCU posts (CE, DE, UC) when command retry
would normally be posted. A subsequent sense command to the SCU sends sense bytes 0, 1, and 2 of hex
04 00 00 (Overrun), to the host system, plus the
format/message and data bytes which correspond to
the specific error condition. The normal recovery for
Overrun is to retry the entire channel program 10 times.
Sense byte responses to specific error conditions are
listed below:
a. Data Check Uncorrectable recovery causes
Overrun and Format 4 to be posted in the
sense data. The data check may be due to a
missing Address Mark or a sync check.
b. Data Check Correctable recovery causes
Overrun and Format 5 to be posted in the
sense data.
c. Seek Check recovery causes an Overrun and
Format 1, Message A to be posted in the sense
data.
Alternate/defective track recovery is modified to bypass
the presentation of the command retry disconnect status
and the reconnect status. The recovery action takes place
during the command execution with no indication to the
channel.
Write offset recovery is not required as no offset for
data checks occurs in the not-RPS mode and offset is
reset when the seek at the start of a channel program is
processed.
Index continue recovery is modified to bypass the
presentation of command retry disconnect status and
reconnect status. The recovery action takes place during
the command execution with no indication to the
channel.

Overrun error recovery causes Overrun and Format 0,


Message 0 to be posted in the sense data.

tion. This status is also given on 2860 channels when


the SCU is not in RPS mode, but no channel disconnect
occurs and the channel remains connected to the SCU
until the mechanical motion is completed and DE status
presented. This 2860 selector channel sequence also
occurs on a 370 channel when the channel program is
executed using a shared unit control word (UCW) in the
channel rather than the normal unshared UCW.

3.12.3 I/O Programming

When the SCU is in RPS mode, the CE-only status is


modified to force a disconnect. The mOdified status is
CE, UC, UE.

3.12.3.1 CHANNEL PROGRAMS

Command retry status (SM, CE, UC) is a special case of


disconnect status.

Track padding disconnect for a format write command is


modified to not issue command retry disconnect status
and reconnect status. The padding takes place during the
execution of the command with no indication to the
channel.

The SCU will execute channel programs which run on an


IBM 3330 drive. Timing differences may occur due to the
simulation of disconnected command chaining.
Command retry requires that the CPU software locate
the CCW which contains the failing command, using
the CCW address supplied by the CSW when the 1/0
interrupt occurs. The failing command is at the CSWindicated address minus eight, except when data chaining. Data chaining may cause the indicated CCW to be
a data address pointer rather than the command CCW.
3.12.3.2 STATUS CONDITION EXCEPTIONS
Status conditions added or changed as a result of the
2860 Attachment Feature are listed below.

Unit Check (UC)


UC status is presented when the SCU cannot continue
a command chain due to an error condition. The data in
a subsequent sense indicates the nature of the error.
The exception to this usage occurs for disconnect
status (CE, UC, UE) on a 2860 channel when the SCU is
in RPS mode and for command retry status (SM, CE,
UC). In these cases, no sense data is required to define
the error.
When disconnect status (CE, UC, UE) or command
retry status (SM, CE, UC) is presented, the SCU will
present DE in a subsequent status to signal the request
for reconnection.
Unit Exception (UE)

Control Unit End (CUE)


a. CUE will be included with stacked status
(except during reconnect sequences).

Disconnect status on a 2860 channel is (CE, UC, UE)


when the SCU is operating in the RPS mode. The
normal disconnect status of CE-only is modified to
force disconnection on a selector channel.

b. CUE will be included with DE for unchained


Seek or Set Sector commands which previously presented CE only (when not in RPS mode
only).

3.12.3.3 I/O INTERRUPTS

c. CUE will not be presented in conjunction with


disconnect status (CE, UC, UE) or command
retry disconnect status (SM, CE, UC).
Channel End (CE)
CE only status is presented to initiate a channel
disconnect on block multiplex channels for Seek and
Set sector commands which require mechanical mo-

The channel performs an I/O interrupt to send device


status to the CPU when a channel program ends or as a
result of a channel poll sequence which obtained status
from a device. If the CPU has channel interrupt disabled,
the channel enters the "interrupt in channel" state for
ending status obtained when a channel program ends.
The channel CSW must be cleared before other
operations can be performed by the channel. This can be
done by a Test I/O instruction to the device address for
which the CSW applies, or by enabling the channel
interrupt.

3-35

An "interrupt in channel" condition does not occur for


an interrupt which is pending for status which is
obtained during a channel poll sequence. The channel
will interrupt the CPU and form the CSW by reseleciing
the device and accepting the device status using a
channel-generated Test I/O if channel interrupt is
enabled. If Start I/O or Halt I/O is issued before the
interrupt is taken. however. the status will remain at the
device and the Start I/O or Halt I/O will be performed.

3672.21-0001-10175

3-36

SECTION 4. MICROPROGRAM
4.1 MICROINSTRUCTION
4.1.1 General
The microprogram controls the operation of the 3672based subsystem and is permanently stored on a flexible
disc. Immediately after power has been applied to the
subsystem. the microprogram is read from the flexible
file. and stored in writable control storage (WCS).
The following paragraph contains a description and
format of a microinstruction. and a description of
decoding and translation of a field within a microinstructi<?n. Also presented is a layout and definition of a
microinstruction word in the microprogram. and a
microprogram subroutine example. In addition. a chart
depicting the various routines within the microprogram is
included.

4.1.2 Microinstruction Word Description


The microprogram controls the hardware of the 3672
SCU through the sequencing of microinstructions. These
microinstructions are divided into fields. Each field
controls a specific function in the hardware. In the SCU.
the data for these microinstructions is stored in WCS.
There are four microinstruction word formats. Each
format. and the fieldS contained therein. activates
specific signal lines which control the logic circuits.
Because of the large number of controls needed within
the SCU. the 34 bits of the microinstruction word are
translated into 57 usable bits. This translation concept is
used to allow a smaller memory word size.

4.1.3 Microinstruction Format

(immediate operand Z). The ALU function is specified by


the F field 0 = F(AZ)
The other register is specified by the C field. It is loaded
with the constant (immediate operand) contained in the X
field (C = Xl.
The loading of the destination registers occurs at register
time of the present cycle. Specifying R03 as a register
designator causes operation upon register RO. R1. R2.
and R3 in parallel. When the destination register of one
bus is designated R03 and that of another bus is RO. R1.
R2. or R3. the bus operation on the Single register will
have its data loaded and the other three registers will be
loaded from the other bus.
4.1.3.2 FORMAT 1 MICROINSTRUCTIONS
This format. shown in Figure 4-1. part b. provides the
ability to set data into a register and also branch on a
machine condition.
The BA field is used to form the low-order 6 bits of the
WCS address register. if the condition specified by the
branch condition field BC is satisfied. The high-order bits
of the address register are unchanged when a branch is
made. which limits Format 1 branches to the present 64word storage block (64-word page). The word at the next
sequential storage address is executed if the branch
condition is not satisfied. Since the instruction read and
execute operations are overlapped. the branch decision is
made on the value of the machine conditions at the end
of the previous microinstruction cycle.
The register specified by the 0 field is loaded at register
time with the result of an ALU operation on the data from
the register specified by the A field and the constant Z
(immediate operand Z). The ALU function is specified by
the F field D = F(A.Z) .

4.1.3.1 FORMAT 0 MICROINSTRUCTIONS


This format provides the ability to transfer data into two
registers with one instruction. It is shown in Figure 4-1.
part a.
One of the registers is specified by the 0 field. It is loaded
with the result of an ALU operation on the data from the
register specified by the A field and the constant Z

4.1.3.3 FORMAT 2 MICROINSTRUCTIONS


This format (Figure 4-1. part c) provides the ability to
perform a control operation. set data into a register. and
also branch on a machine condition. The BA and SA2
fields are used to form the low-order 9 bits of the WCS
address register if the condition specified by the branch
condition field SC is satisfied. The high-order bits of the

address register are unchanged when a branch is taken.


which limits Format 2 branches to the present 512-word
storage block (512-word page). The word at the next
sequential storage address is executed if the branch
condition is not satisfied. The branch decision is made on
the value of the machine conditions at the end of the
previous microinstruction.
The constant Z (immediate operand Z) is loaded at
register time into the register specified by the 0 field (0
Z).

The control field CL is decoded into one of 32 control


functions. Some control functions set and reset latches
in the hardware. and others provide register-to-register
data transfers. The destination registers are loaded at
register time. Specifying RO. R1. R2 or R3 in the 0 field
when the register-to-register transfer specifies register
R03 will cause the D Bus data to be used in place of the
register-to-register data for the 0 field register specified
and the other three registers of the group R03 (RO. R1.
R2. R31 will receive their data from the source register of
the specified register-to-register transfer.
4.1.3.4 FORMAT 3 MICROINSTRUCTIONS
This format (Figure 4-1. part d) provides the ability to
perform an ALU operation on two registers. set data into
a third register. and also branch on a machine condition.
The SA and BA3 fields are used to form the low-order 12
bits of the WCS address register if the condition specified
by the branch condition field BC is satisfied. The word at
the next sequential storage address is executed if the
branch condition is not satisfied. The branch decision is
made on the value of the machine conditions at the end
of the previous Microinstruction cycle.
The register specified by the D field is loaded with the
result of an ALU operation on the data from the register
specified by the A field and the data from the register
specified by the B field. The ALU function is specified by
the F field [0 = F(A.B)].

decoded. dependent upon


microprogram word fields.

the

contents

of

the

4.1.5 Field Translation


The control of the hardware by the microinstruction
fields is accomplished through groups of bits which
control specific hardware functions: Because the SCU
requires 57 hardware controls. but only 34 bits are
available in the microinstruction. the microinstruction
word is translated (expanded). During each machine
cycle. the microinstruction format code is used to
translate the microprogram fields (explicitly and implicit!vi into the proper hardware controls.
The relationship of the hardware controls described
below to the microprogram instruction. is shown in
Figure 4-2.

RA Control (5-Bit Group)


This group specifies the register to be multiplexed to the
A Bus.

RB Control (3-Bit Group)


This group specifies the register to be multiplexed to the
B Bus.

RC Control (3-Bit Group)


This group specifies function to be performed by the ALU.

RD Control (5-Bit Group)


This group is decoded to send a set pulse to the proper
machine register to transfer the 0 Bus contents into that
particular register.

4.1.4 Microinstruction Field Definitions

RF Control (4-Bit Group)

Table 4-1. when used with the word formats explained


on the previous page. shows which controls will be

This group specifies the register to be multiplexed to the


T Bus.

4-1

a. FORMAT 0 MICROINSTRUCTIONS
Word Bit Position

Fields

/ 0

2131

FMT
FIELD

/ 6

10

11

12

13

14 1 15 1 16 1 17 1 18

19

20

Specifies register to be
multiplexed onto A Bus

Specifies register to
be loaded from 0 Bus

21 1 22

23

1 24

/ 25 / 26

Specifies constant used


as the 8 entry to the ALU

27

1 28

29

30

31

32 1 33

Provides constant to be
set into register specified
by C field

Specifies
ALU operation

Specifies register
destination of constant
contained in X field

b. FORMAT 1 MICROINSTRUCTIONS

o 1
o I

Word Bit Position

21

10 1 11

12

13

1 14

1 15

16

17

18

1 19

20

1 21 1 22

23

1 24 1 25 J

26 1 27

28

/ 29 1 30 1 31

FMT
FIELD

..

"

BA

BC
I

..

..

ThIS fIeld replaces the


6 low order bits of the
instruction address register
if the branch condition
specified by the BC field
is satisfied

SpeCIfIes condItIons
which if satisfied, will
cause the microprogram
to branch

Same uses as Format 0

c. FORMAT 2 MICROINSTRUCTIONS
Word Bit Position
Binary Word Format

o /

1 32 1 33

2/31

FMT
FIELD

/5/6

CL

1 10 / 11

12 /

13 1 14 1 15 1 16 1 17 1 18 1 19

..

SpecIfIes control
functions: transfers,
increment, shift,
reset, etc.

20. 1 21 1 22

23 1 24 1 25 1 26

BA2

Same uses as Format 0

27

28

29

30

BC

1 31 1 32 I

33

BA

"

Used with 8A to re p lace


9 low order bits in the
instruction address register
if branch condition specified
by BC field is satisfied.

Same uses as Format 1

d. FORMAT 3 MICROINSTRUCTIONS
Word Bit Position

1 '1

21

1516

18 J

101 11

12 1 1a

14 1 15 1 16

17

1 18 1 19 1 20 1 21 1 22

23 l24 1

25

1 26 J

27

28 1 29 1 30 1 31

1 32 1 33

FMT
FIELD

..
Same uses as Format 0

BA3

BC

BA

I
Specifies register
to be multiplexed
onto B bus

Specifies ALU
operation

Used with BA field


to replace the 12 bits
of the instruction
address register if the
branch condition specified
by the BC field is
satisfied

Same'uses as Format 2

Figure 4-2. Microinstruction Formats

4-2

TABLE 4-1
FIELD
CONTENTS
(HEX)

MICROINSTRUCTION FIELD DEFINITIONS

FIELD NAME

00

GO

CL

BC

G03

B (BRO)

Branch if BRO is set.

C (G03

=R03)

Transfer conlents of R03 Inlo G03

= R03.INC)

Write data from R03 inlo buffer at address specified by BAR


and incremenllhe BAR

01

CI

Gl

+C

CT

Rl

B (BR1)

Branch If BRl is set.

C (BUF

02

CHF

G2

+1

CHC

R2

B (BR2)

Branch if BR2 is set.

C (CC = R23)

Transfer contents of R2 and R3 mto CC register. Reset CC :. 0 latch.

03

BO

G3

BI

CO

B (BR3)

Branch if BR3 is set.

C (DTC = R23)

Transfer contents of R2 and R3 Into OTC (Data Transfer Counter)

04

G5

G5

DO

B (BR4)

Branch if BR4 is set.

C (SR

05

G6

G6

CT

B (BR5)

Branch if BR5 is set.

C (BUF

Write data from R03 into buffer at address specified by BAR pnor
to its being loaded from G3. High BAR IS not modified

SW

MD2

PS

B (BR6)

Branch if BR6 is set.

C (BUF = R03)

Write data from R03 IOta buffer at address specified by BAR BAR IS not modified.

06

= R23)
= R02.LD)

Transfer contents of R23 Into SR Register

B (BR7)
Branch if BR7 is set.
No Operation
C(NONE)
L
CNT
~------~~------r---------~---t----~--------~------~---B--(O--_-O-)------------------------------------------------------1---------------------------------------------------------------1
07

CS

08

CCK

09

SFD

OA

ECR

DB

RWC

RWC

DC

01

SP

Branch if 0 Bus

MCM
SEE ALU
FUNCTION
CODES

-B (0 =0)

FDC
$$$

=zero.

C (R03

No branch if 0 Bus:.: zero.

B (0

= FF)

8ranch if 0 Bus

-B (0

=FF)

No branch if 0 bus

B (CARRY)

= FF (hex).

=FF (hex)

BUF)

Transfe, BUF (buffe' data ,eglsler}10 R03


Buffer storage IS not addressed or read

C (R03 = BUF.INC)

Transfer aUF to R03 and Increment BAR

C (R03 = BUF.LD)

Transfer BUF 10 R03 and then transfer G3 to BAR. High BAR IS nol modified.

C (RESET.INDX)

Reset INT Register bit 4 (Index latch)

Branch if Carry latch is set.

C (CH.FREEZE)

Complete channel transfer sequence In progress. then inhibit funher transfers

00

CT

FOD

-B (CARRY)

No branch If Carry latch is set.

C (RESET.K)

Reset compare latches

DE

CO

CO

-B (COMPARE.EQ)

Branch if A " B latch is set.

C (RESET. CKS)

Reset Check 1 and Check 2 latches

OF

DO

DO

B (COMPARE.HI)

Branch If A ' B latch is set

C (READ.CS)

Load word from WCS Into data register

10

RO

G400 ,

RO

B (INT)

Branch

on

selected

interrupt

true.

Controlled

by

IMK

Register.

C (RESET.DBR)

Reset DBR latch

11

Rl

Rloo

Rl

-B (lNT)

Branch

on

selected

interrupt

false.

Controlled

by

IMK

Register.

12

R2

RWoo

R2

13

R3

R300

R3 .

14

INT

INToo

IMK

15

G4

ROoo

G4

B (CNT = 0)
-B (CNT

= 0)

B (LD.SR.

* +1) LOW

16

G7

G700

G7

17

PS

000

PS

C (D. EVEN)

Generate even panty for the 0 bus for the present micrOinstructIOn

C{LO.SR.*+l)

Load next sequential micrOinstruction address IOta SR.

Branch if counter carry latch is false.

C (BAR

load next sequential microinstruction address into SR. Branch to a


microinstruction in the lower portion of WCS. The branch address is
determined from the branch address fields of the microinstruction.

C (BAR = G3)

NOT USED
B (ILACT)

FO Interface

Branch if counter carry latch is set.

Branch if inline is active.

1)

C (CLOCK.DTC)

Branch If DBR (data buffer ready) is true.

B (DBR)

In

Increment BAR and begin buffer read cycle


Transfer G3 to BAR (B bits)
Check the DTC The DTC counts up If RWC bit 5 IS on, and down If RWC bit 5 isoff.

C (BAR

= G23)

Transfer G2 and G3 to BAR (9 bits)

C (G23

= BAR)

Transfer BAR 'to G2 and G3. Registers GO and Gl a'e unchanged

~-------4--------~------~~SELBY---+--------~--------~-------------------------------------------------------------~--------------------------------------------------------------~

18

GO

19

G2

CHFoo

PS BITS
ON BR
BUS

lA
lB

TR

-B (DBR + INT)

GO

No

+C

If

DBR

or

the

selected

Interrupt

is

true

No branch if channel buffer ready (CBR)

CNT

B (Q.EMPTY)

Branch when channel buffers A and B are available for new data

R03

B (NO. BRANCH)

No branch

C (R23

Load next sequential miCrOinstruction address into SR Branch to a


micrOinstruction in the upper portion of WCS. The branch address IS
determined from the branch address fields of the microinstruction.

C IR23 = CC)

Transfer CC (channel counler) to R2 and R3

C (WRITE.CS)

Transfer contents of data register in FD interface to WCS at


address specified by the address register

B (LD.SR.

* +1) HI

lC

Gl
G3

G3

B (UNCOND)

Unconditional branch to microinstruction


branch address fields of this microinstruction

lE

CNT

TR

B (D. BUS)

Branch to microinstruction address developed by logical OR of


o bus with 8 low order bits developed by branch address fields
of this microinstruction

lF

CP

CP

B (ADR

NBROO .....

Gl

C IR03 = G03)

10

0100

= ADD

= ADD

WITH CARRY DETERMINED BY CARRY LATCH

NOT USED

B (CBR)

= SRI

Branch

ALU FUNCTION CODES:

branch

G2

to

microinstruction

address

address

obtained

from

developed

SR

= SRI

by

Register

Transfer G03 to R03


NOT USED
Transfer SR to R2 and R3. Register RO and Rl are unchanged.

NOT USED

C (SPECIAL)

Expands number of control stalements by using three bits on 0 bus.


Following functions are performed via this control statement:

o BUS
567
+1 0. ADD WITH CARRY
- LOGICAL AND

v
*

0
0

LOGICAL OR
EXCLUSIVE OR

K. = COMPARE
L - SHIFT "A" LEFT

= 000 .
001 .
010 '
011 .
100

Not used
NOI used.
ILACT (set ILACT)
ILACT. OFF (reset ILACT).
Not used

101 . Propagale Select Out


110 . Sets 13th address bit for
ReadlWrite Control Storage.
111 Resets 13th address bit for
ReadlWrite Control Storage.

4-3

INPUT TO
B MUX

FORMAT 0

INPUT TO
F MUX

BRANCH ADRS
TO ADRS REG

MICROPROCESSOR
CONTROL FIELDS

_,,~oo,

MICROINSTRUCTION
FIELDS

P-~~--~~--~--~~~--~--~~'

INPUT TO
B MUX

FORMAT 1

RT Control (4Bit Group)


This group is decoded to send a set pulse to the proper
machine register to transfer the T Bus contents into that
particular register.
RBC Control (5-Bit Group)

This group consists of the microinstruction Z field gated


by the microinstruction format code. RK is gated to the B
Bus with certain formats.

This group specifies which branch condition will control


the next address executed by the Control Unit.

RX Control (8-Bit Group)

RN Control (12-Bit Group)

INPUT TO
F MUX

BRANCH ADRS
TO ADRS REG

RK Control (aBit Group)

This group forms the branch address for the WCS. It is


developed by the BA, BA2, and BA3 fields; and certain
bits of the Address Register.

This group consists of the X field of the microinstruction


format code. RX is gated to the T Bus by the format 0
microinstruction.

MICROPROCESSOR
CONTROL FIELDS

,_,~oo,

MICROINSTRUCTION
FIELDS

FORMAT 2

~~~~-.------~~--~~--~
INPUT TO
B MUX

INPUT TO
F MUX

INPUT TO
B MUX

INPUT TO
F MUX

EQUAL TO HIGH ORDER 6


BITS OF ADRS REG

BRANCH ADRS
TO ADRS REG

MICROPROCESSOR
CONTROL FIELDS

TRANSLATION

MICROINSTRUCTION
FIELDS

FORMAT 3

EQUAL TO HIGH ORDER 3


BITS OF ALU

BRANCH ADRS
TO ADRS REG

MICROPROCESSOR
CONTROL FIELDS

TRANSLATION

MICROINSTRUCTION
FIELDS

Figure 4-2. Microinstruction Expansion

4-4

4.2 MICROPROGRAM

3. Starting the diagnostic monitor when sequence 1 and


2 have not been in progress for 150 milliseconds.

4.2.1 Microprogram Block Format


The microinstructions listed on a microprogram page are
formatted into blocks. The format of each block depends
on the format of the microinstruction. The different
formats (0, 1, 2 and 3) plus the subroutine block format
are shown in Figure 4-3. This figure depicts the
difference between each format, and defines the entries
contained in each microinstruction block.

4.2.2 Example Microprogram


The example shown in Figure 4-4 is the actual Read
Home Address routine portion of the microprogram. The
example depicts a sequence of microinstructions and an
explanation of the entries within a microinstruction.

4.2.3 Microprogram Routine Organization


The microprogram IS divided into a number of routines.
Each routine performs a basic function, but may accomplish several different operations, depending on how
it is used.
The microprogram routine organization is shown in
Figure 4-5. A brief explanation of each block is also
contained in the following paragraphs.

Reset Procedure
The machine hardware is Initially reset to the power-on
reset state. The microprogram is then loaded (IMPL)
starting at the IMPL address, or restarted (Selective
Reset and System Reset) at location 000. IMPL starts a
new microprogram with the Writable Co'ntrol Storage
(WCS) and buffer storage in the reset state. General
Reset only resets the devices allocated to the channel
initiating the General Reset command. Selective Reset
only resets the particular operation in progess.

device Interrupt is received and reset when the channel


is acting like a selector channel, and that the new
command is received from the channel.

bytes fa: '1 equal, high, or high equal state. The section
then transfers the status of the comparison to the
channel

File Status Analysis


File Status Analysis sets up the current device
parameters in scratch pages 7-B, when selection occurs.
File Status Analysis also receives the command from the
channel. The parameters and the channel registers are
examined to determine what status should be presented
in ending status. If no ending status exists, the decode of
the Channel Command is entered.

Unchained End Procedure

Write Commands

Unchained End Procedure is used when the channel or


SCU ends a command chain, and the Controller and drive
must be deselected. When Unit Check is presented in the
preceding status, the section must assemble the sense
bytes in Buffer Storage location 120 (hex). The drive
parameters are also stored in the buffer. If In lines are
active, the program loops back to the Diagnostic Monitor
section, otherwise the program goes back to the Basic
Wait Loop section.

The Write Command section accepts bytes from the


channel. and using hardware channel transfer, sends the
bytes to the Controller to be written. The ECC is
appended to the data train to be written by the Controller.

Command Decode
Command Decode can be entered from File Status
Analysis, or from Chained End Procedure. When entering
from Chained End Procedure, the time through command
decode is taken up in the gaps. Command Decode checks
the validity of the command, checks that the proper
sequencing has occurred prior to execution of the
command, and sets up the command indicators required
to process the command.

Initial Status Presentation


Initial Status Presentation presents zero status, for the
SCU commands, to the channel. If orientation exists with
the track, the routine will wait for Command Out Signal
to drop, raise Status In, and then exit. If orientation does
not exist, the sequence is completed by waiting for
Service Out in response to Status In.

Sense Commands
Four Sense Commands accept sense information from
various locations in the Buffer Storage and send the
information to the channel for analysis. A hardware
channel transfer is used to switch DATA IN and SERVICE
IN when transferring the sense bytes.

Control Commands
The Control Command section is not directly involved
with transferrring data from the 3670 file. This section
moves the access position, reads and sets the sector
values, receives the file mask from the channel, and
spaces across bad count fields.

Read Commands

Ending Status Presentation presents ending status to the


channel. The section then waits for channel acceptance,
and determines if the channel wants to continue the
chain, discontinue the chain, or block multiplex.

The Read Command section transfers data directly from


the Controller, using hardware channel transfer. Correctable errors occurring in the HA. Count, or Key Fields are
corrected in Buffer Storage. The corrected data is
transferred from the Buffer Storage using hardware
channel transfer.

The Basic Wait Loop responds to any of the following


sequences:

Chained End Procedure

Search Commands

1. Initial Channel Selection


2. Control Unit Initial Selection

Chained End Procedure is used when the channel chains


one command to another. This section insures that the

Search Commands section accepts bytes from the Controller and from the channel. The section compares the

Ending Status Presentation

Basic Wait Loop

Diagnostic Commands
The Diagnostic Commands section controls the in-line
microdiagnostics from the CPU. The section also enables
the 367X Fault Isolation Detection System to evaluate
the test results.

Error Processing
The Error Processing section handles all retryable errors
and error analysis. The retryable errors are: (1) ECC
errors, (2) Seek Incomplete, (3) Defective and Alternate
track errors, (4) Retry pading, (5) Sync check, (6) AM
check, (7) Index Continue, (8) Offset before Write, (9)
Command Overrun, and (10) Data Overrun.

Diagnostic Monitor
The util.ization of all microdiagnostic tests is accomplished through the resident SCU Diagnostic Monitor
section of the Microprogram. The Diagnostic Monitor
interfaces with the functional Microprogram and with the
microdiagnostic required to perform a specific test.

Microdiagnostic
The Microdiagnostic routines are loaded into the CPU
from the Flexible Disc, and then transferred to the WCS.
Execution of a microdiagnostic is initiated by a Diagnostic
Write Command. The microdiagnostics test the SCU,
Controller, and the Disc Drive sequences; and monitors
any errors which may occur. The errors are stored in the
sense bytes for analysis by the Error Recovery Procedure
(EREP) or the Field Engineer (FE).

4-5

BLOCK IDENTIFIER
LOCATION OF

SAME AS FORMAT 0

FORMAT 1

MICROINSTRUCTION

SUBROUTINE BLOCK

SUBROUTINE

NAME

--RETURN

1
ADDER]

~~

L: :

A-:-:-:-D-N--TT-I-O-N

ADD

A~~

ALWAYS ON LINE

4
BRAN CH
ADDRESS
LABE
B L K - - LIN E O P - - - BRA D D

BLK-LI

NE-B-ADDR

8
9

OF

SUBROUT IN

' " O'NO'" ,u",OU"N' ",0"


SAME AS

FORMAT 0

SAME AS FORMAT 0

ALWAYS CODED IN LINE 4


BLOCK IDENTIFIER
/
BLOCK IDENTIFIER
ALU STATEMENT MAY
\
USE LINES 1,2 AND 3
IF REQUIRED. OTHER
WISE LINE 2 MUST BE
USED.
0

LOCATION OF

FORMAT 0
LAB E L
A

S TAT E MEN T

TBU

STATEMENT

~~TRR6~N~~~~6i~~..N.O,_F_ _ _6

B L K --

ON A PAGE
17 CHARACTERS
MAXIMUM

COMMENTS

R/~:::::::::N'

3
d 0

3
4

LOCATION OF
MICROINSTRUCTION

FORMAT 2

IS ALWAYS ON
LINE 4

ALU MAY USE LINES 1 AND 2


IF REQUIRED, OTHERWISE
LINE 2 MUST BE USED.

1
2

A::E:TATEMENT

;~~~~~~~T~(~~~N._T~IS~____-4~~~ CON T R 0
5
7

LIN E - - OP- BRA D D R


\

o
AD;}R

S TAT E MEN T
CONDIT ION
ADDRESS
ABEL

BRANCH
BRANCH

~~~~~~

OR

~AOOA

8
",-",ANOHAOOR'"
SAME AS FORMAT 0

FOR ENGINEERING
PURPOSES ONLY

OPERATION CODE.
DENOTES FORMAT
0,1, 2,OR 3.
SAME AS FORMAT 0

RELATIVE LOCATION OF MICROINSTRUCTION WITH RESPECT TO THE BEGINNING OF THE ROUTINE.


ABE L

ADD R

A L U
ALWAYS ON LINE 4

ALWAYS ON LINE 5

5
6
7

8
9

Figure

S TAT E MEN T

3
4

B, R A. N. C H
B R A N C H
L K--L I
0 M MEN

J
/

FORMAT 3

LOCATION OF
MICROINSTRUCTION

STATEME,NT
ADDRESS

LABEL

~:~OOR
-

SAME AS FORMAT 0

4-3. Microprogram Block Formats

4-6

Resets bit #3 in Rl page B

Sets PS Reg to page Band


BR MPX to NBR

Clears G5 Reg by AN Dong

by andmg R1 with constant

EF

GO with

'~

r-----~..........

Indicates a Subroutine

7f~ro

Lrl

' - r - - - - - 003

Rl=Rl. /EF

;--------i

G5=O.GO
B(LO.SR. "1)

O4AINOEXl
B3--011-3--ooo
CLR ERROR REG.
GO GET INDEX
ORIENTATION.

B2--OO8-1-- 003
RESET INDEX
PASSED IN OP STAT 1.

I N O E x . l - ! - 004
SUBROUTINE O4A

84--014-B

Loads SR Reg with return


address and branches to
subroutine
Adds Hex 08 to G 1 reg and
puts result into GO reg

Clears G5 Reg by ANDing


GO with zero
~--

'

--004

..

~i _ _ _ _ _ _

~
-8(0=0)
ERROR
01-015 3 - - 000
BR IF SUBRTN ERROR

(:

Indicates a Subroutine.
"'HA\

r-----+-OO~

Sets IMK reg equal to hex

lB

Branches to itself until CNT


reg decrements to zero.
006

G50 G
:G
- s - 1 SUBROUTINE 05F
:
0

B(LO.SR..I)
05FHA
02---017-3
CLR ERROR REG.
GO TO LOAD CNTS

Branch If Hl-e pr~vtOUS D


Bus was not equal to zero

-B(CNT=O)
.

03-019-B

'/0
05-022--0-- 000
SET GO FOR CHNL
XFER IN 040.
"'AOJ TIMER'"

Branch If the prevIous 0


Bus was not equal to zero

06--025-2-SET MASK 0 CNT=O.


INDEX .RESP.CK2
WAIT FOR 30 /lSEC
Resets Rl reg of page B by
anding G2 wIth zero

Set PS Reg to page B and


BR MPX to NBR

,------009

r-----7<-----DOA

Branch if the previous 0


Bus was not equal to zero

Indicates a Subroutine

ERROR

Loads SR Reg with return


address and branches to
subroutine
~ooo
.
ERROR.ASMB--OOI

SUBROUTINE 04F
B(LO.SR. "1)
O4FERROR.ASMB
H2--OO3-3
ASSEMBLE ERRORS.

H3--OO5-B

Figure 4-4. Microprogram Format Example

4-7

t----<

r--t-'

SENSE
COMMANDS

I-- t--<

r-

r--

DIAGNOSTIC
MONITOR

I
I

r-----

r----<

MICRO
DIAGNOSTICS

CONTROL
COMMANDS

1
~

READ
COMMANDS

t--<

L.-..

BASIC
WAIT
LOOP

f-----

FILE
STATUS
ANALYSIS

COMMAND
DECODE

r-----

INITIAL
STATUS
PR ESENTATION

r-t-- r

--~

I--- 1--1

..........

I-- r-

I-SEARCH
COMMANDS

r-- ~

J
~

GENERAL
RESET

SELECTIVE
RESET

WRITE
COMMANDS

I-- ~

RESET
PROCEDURE

UNCHAINED
END
PROCEDURE

ERROR
PROCESSING

ENDING
STATUS
PRESENTATION

CHAINED
END
PROCEDURE

I""-

I-DIAGNOSTIC
COMMANDS

IMPL

J
Figure 4-5. Microprogram Routine Organization

4-8

SECTION 5. FLEXIBLE DISC FILE

5.1 GENERAL DESCRIPTION


The Memorex 650 Flexible Disc File (Figure 5-1) is a
compact, direct access, removable disc unit intended to
simplify the distribution, processing, and storage of
digital information. The 650 stores the subsystem
microprogram and the microdiagnostic program library.
The 650 is composed of a drive mechanism, read head,
head actuator, and associated electronics. Connections
are made to the SCU Microprocessor Flexible Disc

Interface by signal and power cables, which supply


addressing, function requests, data formatting, and
power.
The FD/"I Flexible Disc Cartridge (Figure 5-2) used in the
650 is a flat disc composed of a Mylar substrate coated
with a magnetic oxide. For protective purposes during
handling, operation, and storage, the disc is permanently
encased in a flexible plastic jacket, 8 inches square by VI6
inch thick.

Oc=)
JACKET

Figure

1 Memorex 650 Flexible Disc File

Figure 5-2. Flexible Disc Cartridge

5-1

5.2 SPECIFICATIONS

5.2.1 Machine Characteristics

Machine characteristics include data retrieval time, disc


characteristics, and data recording format. These items
are described as follows.

Record Length Sectorized


(8 per track)
Record Length Indexed
(1 per track)
Disc Capacity Sectorized
Disc Capacity Indexed

tion. All maintenance and diagnostic procedures are


predicated on the proper application of power from the
SCU flexible disc interface.

3.5 kilobits
30 kilobits
1.4 megabits
1.5 megabits

Data Recording Format


Data Retrieval Times

Rotational Speed
Single Track Access Time
Data Transfer Rate

375 rpm
20 msec
200 kilobits/sec

Disc Characteristics

Number of Tracks
Recording Density

50
2400 bits per inch
(inside track)

Recording Mode

Frequency
tion

Sectors per Track


Index per Track

8
1

modula-

All 650's are capable of being operated with 50 or 60 Hz


power. For 60 Hz, the disc drive motor pulley is mounted
on the shaft with the large end toward the motor. The
belt is driven by the small diameter pulley. For 50 Hz, the
pulley is reversed on the shaft and the belt is driven by
the large diameter pulley.

5.2.2 Operating Capabilities

5.2.3 Power Requirements

Operation of the 650 in the SCU is fully automated,


requiring no operator intervention during normal opera-

All power required by the 650 is supplied by the SCU


flexible disc interface.

AC Power
110 10% volts
50/60 Hz, single phase
0.75 amperes

DC Power
+5 0.10 volts @ 0.6 amps, 50 mV ripple
-15 0.30 volts @ 0.12 amps, 50 mV ripple
*-12 0.25 volts @ 0.12 amps, 50 mV ripple
+24 1 volts @ 2.0 amps, 100 mV ripple

May be used in lieu of -15 volts with no modifications


to the file or cables.

5.3 SPECIAL PRECAUTIONS

5.3.2 Disc Interchangeability

The 650 can be damaged by improper servicing, handling, or operating techniques. The following procedures
should be observed to properly operate and maintain the
650.

To ensure interchangeability, store discs in a location


that is within 5F of the using system ambient
temperature and within 10% of the using system
humidity. Discs stored outside the recommended ranges
must be placed in the uSing system environment at least
20 minutes before use.

5.3.1 Cartridge Loading


The cartridge consists of the flexible disc encased in a
plastic jacket. Wipe cushions are bonded to the inside of
the jacket. The disc is housed and rotates between these
cushions during normal operation. Figure 5-3 shows how
the cartridge is loaded in the cartridge guide. To load the
cartridge open the door, insert the cartridge into the
cartridge guide, and close the door.

1. Return the cartridge to its storage envelope whenever


it is removed from file.

6. Do not smoke while handling cartridges. Heat and


contamination from a carelessly dropped ash can
damage the disc.

2. Store cartridges vertically.

7. Do not expose cartridges to heat or sunlight. The


3. Keep cartridges away from magnetic fields and from
ferromagnetic materials which
might become
magnetized. Strong magnetic fields greater that 50
oersteds can distort recorded data on the disc.

Read/Write head on the 650 cannot properly track a


warped disc.

8. Do not touch or attempt to clean the disc surface.


Abrasions may cause loss of stored data.

5.3.3 Physical Damage


When removed from the 650, the disc cartridge is stored
in a plastic-coated paper envelope (Part No. 204268). To
protect the cartridge, the same care and handling
procedures specified for computer magnetic tapes apply.
Additional precautionary procedures are as follows:

4. Replace storage envelopes when they become worn,


cracked, or distorted. Envelopes are designed to
protect the disc.
5. Do not write on the plastic cartridge. Writing pressure
may damage the disc.

5.3.4 Safety
AC and dc power are controlled by the SCU. Before
working on the file, verify that all power is removed from
the 650.

KEYING
NOTCH

CARTRIDGE
GUIDE

Figure 5-3. Cartridge Loading

5-3

5.4 PRINCIPLES OF OPERATION


5.4.1 General Operation
The 650 consists of control and read electronics, disc
drive motor, read head, head position actuator, and
removable disc (Figure 5-4). The primary functions
performed by the 650 are:

Sector/Index Detector Circuits


Track Position Actuator Driver
Head Load Actuator Driver
Read Amplifier and Transition Detector
Data/Clock Separation Circuits

An electrical stepping motor (Head Position Actuator) and


lead screw position the read head. Step in and step out
pulses from the flexible disc interface cause the stepping
motor to rotate the lead screw clockwise or
counterclockwise in 15-degree increments. A 15-degree
rotation of the lead screw moves the read head one track
position.
The Disc Drive Motor rotates the spindle at 375 rpm
through a belt-drive system. Either 50 or 60 Hz power is
accommodated by means of a stepped pulley. A registration hub, centered on the face of the spindle, positions
the disc. A clamp (that moves in conjunction with the
door) fixes the disc to the registration hub. The disc is
held verti(;allv, and rotates in a plastic jacket, which
protects and cleans the recording surface during operation.

.....
/

I
I
I
I
I

.... CARRIAGE AND READ


HEAD ASSEMBL Y

READ HEAD

I
SCU
MICRO
PROCESSOR

.....

FLEXIBLE
DISC
INTERFACE

The Control and Read Logic interface directly with the


flexible disc interface electronics in the SCU
microprocessor, described in paragraph 6.3.6. The Head
Positioning Actuator positions the read head at the
desired track on the disc and the Head Load Actuator
loads the disc against the head, so that data may then be
read from the disc.

The electronics are packaged on one PCB. The PCB


contains:

Receive and generate control signals


Generate status signals
Access the appropriate track
Read data upon command

CONTROL,
STATUS,
AND
READ
LOGIC

LOAD HEAD

HEAD
POSITION
ACTUATOR

I
I
I

I
HEAD POSITION ACTUATOR CONTROL

I
650 FLEXIBLE DISC FILE

-~

Figure 5-4. Flexible Disc File/Microprocessor Interface

5-4

The disc, 75 Inches In diameter. has eight holes spaced


around the perrphery for sector definition There IS also
one additional hole for Indexing. The disc rotates Inside
the plastic envelope during normal operation Figure 5-7
Illustrates the disc and disc envelope

5.4.2 Functional Assemblies


The 650 IS divided into three functional logic blocks and
electromechanical transducer as shown in Figure 5-5.
The read head is mounted on a carriage that is moved by
the lead screw. Head loading is achieved when the disc
is lightly loaded against the rigidly mounted head by
moving a load pad against the disc with a solenoid
actuated bail. Head to disc compliance is achieved by
restraining the disc between the head the load pad.
Figure 5-6 schematically illustrates this action.

SOLENOID
(ENERGIZEDI

LOAD ARM

The Light Emitting Diode (LED) and Detector generate


disc index and sector pulses used by the Microprocessor
to format and orient data written on the disc. As shown
in Figure 5-7, eight holes (0.10 inches diameter) divide
the disc Into equal sectors. The ninth hole (same
diameter), spaced midway between two sector holes.

RETURN
SPRING

CARRIAGE

READ/WRITE H E A D " DISC

I
I

'

---=:J~~~_

SEP DATA

READ HEAD

-SEP CLOCK
DETECTOR

Figure 5-6. Head Loading Mechanical Assembly

-STEP OUT

INDEX/SECTOR
ACCESS HOLE

-STEP IN

TRACK 00
TRACK 00

-LOAD HEAD
-SECTOR
-INDEX
-TRACK "00"

I
I

COI\JTR.OL
LOGIC:

-LOAD HEAD

-VAllO LO-'lD HEA8

CARTRIDGE
DOOR SECURED

HEAD POSITION ACTUATOR STEP CONTROL,

<;

HEAD POSITION ACTUATOR STEP CONTROL,

HEAD POSITION ACTUATOR STEP CONTROL.

I
I

READ/WRITE
HEAD OPENING

TRACK 49
OPENING FOR
DRIVING HUB

COMPOSITE INDEX/SECTOR

"-_ _ _ _ _....:..;:..:.._ _ _ _ _ _r - - PLASTIC


JACKET

Figure 5-5. Flexible Disc File Functional Block Diagram

Figure 5-7. Disc Cartridge and Disc Configuration

5-5

indicates one disc revolution. The LED and Detector


(photo transistor) are placed on opposite sides of the disc.
As the disc revolves, the holes pass between the LED
and Detector, exposing the Detector to infrared light,
which turns on the the Detector. The Detector output is
shaped by a threshold detector, and an output pulse is
obtained. Output is approximately at 2.5 volts with a
transition to 5.0 volts for the pulse.

5.4.3 Logic And Analog


Functional Descriptions
These paragraphs discuss each logic block and signal
name. The descriptions are divided into two logic blocks
Control and Read. All input interface functions are
generated by the SCU. All output interface functions are
generated by the 650.

5.4.3.1 CONTROL AND STATUS LOGIC

Transducer Signals between the 650 Control Logic and


electromechanical assemblies are as follows:
HEAD LOAD

Energizes the solenoid as shown


in
the
head-loading
block
diagram. It is a function of LOAD
HEAD.

TRACK 00

Switch closure indicating that


the read head is located at Track
00.

VALID LOAD HEAD

Switch closure interlock indicating that the cartridge door


is secured. If this condition is
not satisfied, the read head cannot be loaded.

LED

COMPOSITE
INDEX/SECTOR

The basic functions of the Control and Status Logic are to


place the read head on the proper track, hold the disc
against the head for read operations, and indicate disc
rotational position. The SCU interface functions are as
follows:
STEP OUT

Increments the Head Position


Actuator. Each pulse moves the
head outward one track away
from the center of the disc.

STEP IN

Increments the Head Position


Actuator. Each pulse moves the
head one track inward toward
the center of the disc.

LOAD HEAD

Loads the disc to the read head.

SECTOR and INDEX

Provides disc sector and index


position information. Each signal
is a separate output. These
pulses are generated from holes
located around the periphery of
the disc and are used for the
accessing of data on the disc.

TRACK 00

Indicates when the read head is


located at Track 00.

Provides power to the


Emitting diode (LED).

Light

Detector signal input to the Control Logic providing disc sector


and index information. Within
the Control Logic, the INDEX
and SECTOR signals are
separated, and are transmitted
by separate interface lines to
the SCU.

READ HEAD

An analog representation of the


recorded information consisting
of composite data and clock
signals, differentiated by alternating polarity. Those Signals
are amplified, differentiated,
limited, and then shaped. The
shaped data is applied to a data
separator
circuit
which
separates the data bits from the
clock bits.

5.4.4 Functional Operation


The 650 functional sequences are divided into the
following three phases of operation: initialization, track
access, and read. The initialization phase is used whEm
the power is turned on. During normal operation, the
head is pOSitioned at the addressed track during the track
access phase, and the read phase is performed.
5.4.4.1 INITIALIZATION PHASE
Whenever the 3671 Storage Control Unit applies power
to the 650, a Power On sequence occurs automatically to

prepare the disc file for operation. The events and timing
for this sequence are shown in Figure 5-8.
The SCU flexible disc interface applies primary and
secondary power to the 650.
After a two-second
delay, -STEP OUT is pulsed until the head is positioned
at Track 00. This operation is performed to ensure that
the head is properly oriented before a read operation
begins. When the head is positioned at Track 00, the
TRACK 00 signal becomes true. It changes to false when
the head leaves that track.
The -LOAD HEAD signal can be applied any time after
the power has been turned on. When the -LOAD HEAD
line is false, the head is unloaded from the disc. The disc
must be loaded on the head before a read operation can
begin.
5.4.4.2 TRACK ACCESS PHASE
The -STEP OUT and -STEP IN functions are used for
positioning the head to the desired track. The only
restriction placed on these Signals is that each pulse
must be spaced by at least 20 msec. The -STEP IN pulse
width is the same as the pulse width for -STEP OUT.

PRIMARY
POWER (AC)

SECONDARY
POWER (DC)

5.4.3.2 READ LOGIC


The basic function of the Read Logic is to receive analog
signals from the disc and convert this composite signal
into separate clock and data pulses. Data and clock come
from the Read Logic block when the disc is loaded onto
the head. SCU interface functions are as follows:
SEP DATA

Digital data bits read from the


disc.

SEP CLOCK

Digital clock bits read from the


disc.

Transducer signal communication between the 650 Read


Logic and the read head is as follows:

STEP OUT

----1

I-

2.0 SEC MIN

.-jl

- - -.....

""1.
.
- - -

T~: :sEE:

---------------2-0-.0-M-S-E-C-M-IN-------------~~~rf1~---------------..

TRACK 00

LOAD HEAD

-------------------------~frJ
r --- -------------

Figure 5-8. Initialization Sequence

5-6

5.4.4.3 READ PHASE


A read operation comprises a track address and index. /sector seek sequence followed by a read data sequence.
Waveforms for a read sequence are shown in Figure 5-9.
The Read logic is shown in the Logic Manual on page
FD310. The Read head reads the combined clock and
data pulses recorded on the disc. The read logic amplifies
these signals and separates them into two outputs:
separated clock signals (SEP CLOCK) and separated data
signals (SEP DATA). These waveforms are shown in
Figure 5-10.

connector. polarity. logic level. and pulse width. The


wiring diagram of the 650 with signal names. pin
numbers. and connectors is shown in Figure 5-11
The 650 requires only two cables. power and control/data. The twisted-pair lines are physically four
wires. and are wired as illustrated below.
SEP CLOCK

~0.25 jiS (TYPICAL)


SEP DATA

5.4.5 Interface Signal Description

--..l

:'-2.5J1Sj

The interface signals can be divided into three


categories: control. data. and power. The following
paragraphs describe the signals with relationship to

LO.25I1S (TYPICAL)

READ
DATA

STEP IN/OUT

I_
.....J

110.0 IlSEC TO 10.0 MSEC

1---20.0 MSEC

~~~_ _.....

I
MIN~

+5.25 V MAX

1 - - - 2 0 . 0 MSEC MIN
READ ENABLE

j ~.,.-----

+3.3 V MIN
+0.4 V MAX

'--______---f,s

VALID
READ DATA

--f

~ 1.0 MSEC

U~-------'ll~~
-.l
14--2-50--~1
100 NSEC

+100 NSEC

n'-_____

SECTOR

Jl

INDEX

_________________________
10~.~0~M~S~EC~_n.....

I---- 20.0 MSEC

+0.0 V MIN

------L-

--l I-- 1.0

I-

Figure 5-9. Read Sequence

-50 NSEC
50 ns

5.0 jiSEC 0.5% ---------~

MSEC
5r\--------~nL

160.0 MSEC

-I

Figure 5-10. Read Data Waveforms

5-7

5.4.5.1 CONTROL

HOST
SYSTEM

FLEXIBLE
DISC
FilE

COMMOIII RETURIII

-LOAD HEAD

'-'"\

COMMOIII RETURIII

AX

Input Signals
Table 5-1 lists characteristics of the input Signals.
Output Signals
Table 5-2 lists characteristics of the output signals.

10
11

TABLE 5-1.

INPUT CONTROL SIGNALS

12

'-'J'""'i

~INDEX

16

f"'\X
X'-'

COMMON RETURI\j

- TRACK 00

INPUT SIGNAL
DESIGNATION

17

CONNECTOR
AND PIN

ACTIVATION
POLARITY

PULSE
WIDTH

LOGIC
CARD

-SEP DATA (COAX!

-STEP IN

J1-1

Negative

COMMON

Jl-2

Negative

10 /lsec to
10 msec

r...

,)"

,.....

6
7

LOGIC GROUND
5 vDC

D.C.
POWER
SUPPLIES

r\.

X
'-"

-15 VDC 1-12 VDCI

A
Y

~r-v

50/60 HZ

DEIIIOTESSHIELDEDWIRE

~
MOTOR

-.LOAD HEAD

J1-10

Negative

COMMON

Jl-11

Negative

head load

Return

solenoid

TABLE 5-2_

J2

yy
-.0

Negative

FRAME GROUIIID

Jl-3

10 /lsec to

Enables

Level

50 OR 60 HZ

110 V

-STEP OUT

RETURN

42

r...

A.C. POWER

Positioning

Return

RETURN

10 msec

32

X
'-"

-+24 VDC RETURN

J;

24

+24 VDC

Track

-5EP CLOCK ICOAX,

L--

COMMENTS

18

'-'

r-'\.

The interface control Signals are divided into two types:


input and output. The input Signals are provided by the
SCU flexible disc interface, and output Signals are from
the 650. Paragraph 5.4.3.1 previously described these
signals. Logic levels vary, depending on the particular
signal requirements. Signal levels are indicated on the
logic diagrams in the Logic Manual. Negative logic,
however, is used throughout the system. Logical 1 is
true, and is a low level signal, indicated by a minus sign.

~'-'

-SECTOR

FUNCTIONS

1
2

X'-'

-STEP OUT

COIIITROL AIIID
DATA

J1

r...
r...X

-STEP 1111

Logical 0 is false, and is a high level signal, indicated by


a plus sign.

OUTPUT CONTROL SIGNALS

CONNECTOR
AND PIN

ACTIVATION
POLARITY

PULSE
WIDTH

-SECTOR

Jl-12

Negative

1.0 msec

COMMON

Jl-l1

Negative

-INDEX

Jl-16

Negative

COMMON

Jl-17

Negative

-TRACK 00

Jl-18

Negative

COMMON

Jl-17

Negative

OUTPUT SIGNAL
DESIGNATION

COMMENTS
Indicates location
on disc

Return

RETURN

1.0 msec

Indicates location

DEIIIOTESTWISTEDPAIR

Return

RETURN

Figure 5-11. Flexible Disc File Interface Diagram

RETURN

on disc

Level

Indicates when head


is positioned on Track 00

Return

5-8

5.4.5.2 DATA

5.4.5.3 POWER

Data input and output signals characteristics are listed in


Table 5-3.

Table 5-4 lists the power requirements

TABLE 5-4. POWER REQUIREMENTS

TABLE 53

DATA SIGNALS

POWER SIGNAL CONNECTOR


ACTIVATION
DRIVE
DESIGNATION
AND PIN
'
POLARITY
CHARACTERISTIC'
DATA SIGNAL
DESIGNATION

CONNECTOR
AND PIN

ACTIVATION

PULSE

POLARITY

WIDTH

LOGIC
COMMENTS

J15

Negative

0.25 j.lsec

Output data from disc

J1-6

Negative

0.25 j.lsec

Output clock from disc

DATA

-SEP

-15VDC

Logic ground

+5: 0.10VDC

Power

Logic Power

@0.6A

level

supply

50 mv ripple

@0.12A

Power

for read/write

50 mv ripple

level

amplifiers

ground
Jl32

Jl42

Positive

I!
I

I
I

(-12VDC)'

ground
I

DC power supply

-15 0.30VDC

Negative

CLOCK

-12VDC can be

-.12 0.25VDC
@0.12A

+24VDC

Jl8

I!
I

Positive

+24VDC

Jl9

RETURN

I
I
I

, used in lieu

i of

50 mv ripple
+22 lVDC

Power

@2.0A

level

DC power

ground

ground

ground

Line

50/60Hz

Three

AC

terminal
socket

for head positioning

solenoid

DC power

J2

-15VDC

DC power supply

motor and head load

100 mv ripple

DC Power!

110VAC

COMMENTS
DC power supply

Logic ground

Logic

GROUND
+5VDC

-SEP

Jl24

PULSE
WIDTH

110

+24VDC power

ground

!Must be provided

10"10 VAC

@0.75A

Line

50/60 t 0.5Hz

AC

single phase

! from a branch
circuit protected
at no more than
20 amperes.

FRAME

J2

Frame

Frame

Frame

Center socket

GROUND

Center

ground

ground

ground

of 3wire AC

Socket

socket

5-9

5.5 CONTROL CIRCUITRY DESCRIPTION

5.5.1 Functional Description


The electrical and electronic circuitry consists of control
and status circuits composed primarily of solid-state
integrated circuit components mounted on a single
printed circuit board (PCB). The block diagram of Figure
5-12 shows the primary motor, actuators, switches,
circuit components and circuit proups, and the
associated control and status signals. The circuitry
involved in the generation and transfer of the control and
status signals is shown on the schematic and logic
diagrams in the Logic Manual.

5.5.2 Printed Circuit Board


The PCB (Figure 5-13) is mounted at the rear of the 650
enclosure. It is secured to the baseplate by four screws,
one in each corner. A connector plug on one side
connects with connector PC1. The PCB drawing shows
the component side of the board, and indicates the
locations of the test points. Refer to the Logic Manual,
pages FD300 through FD320, for test point circuit
locations and normal waveforms and test values.

5.5.3 Harness Assembly


The harness assembly comprises connector block J1, PC
board connector housing PC1, and the interconnecting
cable consisting of the primary 650 input/output leads.

LIGHT
SOURCE
DISC

DETECTOR

Figure 5-13. Flexible Disc File PCB

INDEX

TRACK 00

POWER
SECTOR

SEP
CLOCK

LD HD

IN

OUT
STEP

Figure 5-12. Control Circuitry Block Diagram

5-10

SECTION 6. STORAGE CONTROL UNIT DESCRIPTION

6.1 GENERAL DESCRIPTION


The 3672 Storage Control Unit (SCU) interfaces between
the using IBM System, the 3673 Controller, and the Disc
Drive Modules. The SCU contains the hardware used to
control operation of the 3672-based Subsystem. The
SCU is organized into four major hardware areas:

Channel Interface -

Communications link
between the SCU and the
using IBM system.

Microprocessor

Contains the microprogram,


ALU, special purpose
registers, error checking circuits, Flexible Disc Interface,
and data busses needed to
control the Subsystem.

Controller Interface-

Communications link
between the SCU, the Controller, and the DDM's.

FE Interface

Contains the logic required


to manually operate and control the FE panel for
diagnostic
control
and
maintenance .

The relationship of these areas to each other is shown in


Figure 6-1.

The microprogram generates control signals for the SCU


during command execution. Microprogram subroutines
and branching conditions allow the SCU to perform
multiple operations, such as reading from one disc file

Operation of the SCU is directly controlled by a


microprogram stored in the Writable Control Storage
(WCS) portion of the Microprocessor.

FROM
USING
SYSTEM

PROVIDES INTERFACE
MICROPROCESSOR.

COUNTS NUMBER OF BYTES IN DATA TRANSFER OPERATIONS

DECODES ADDRESS FROM USING SYSTEMS

ESTABLISHES PRIORITY FOR MULTICHANNEL OPERATIONS

Controller and disc drive status is monitored by the


microprogram for examination by the using channel, and
stored in the Buffer Storage section of the
Microprocessor. Data integrity is constantly checked by
the SCU through the use of parity checking and Error
Correction Codes.
The FE Interface section works in conjunction with the
FE Panel to provide a means for the FE to manually
control the Subsystem for maintenance and
troubleshooting. The capability for running inline and
offline diagnostic routines is also contained on the FE
"Panel.

BETWEEN

THE

650

FLEXIBLE

DISC

AND

THE

MICROPROCESSOR

COMMUNICATES WITH USING SYSTEM CHANNEL (COMMANDS. DATA. AND


STATUS)

while completing a Seek operation on another disc file.


The control Signals generated by the microprogram cause
the Controller Interface to send orders to the DDM's, via
the controller, enabling data to be stored or retrieved.

FLEXIBLE DISC INTERFACE

CHANNEL INTERFACE

The Flexible Disc File provides permanent storage of the


microprogram as well as offline and inline diagnostic
routines. During an initial Power On sequence of the
Subsystem, the microprogram is read from the Flexible
Disc and stored in the WCS. The Flexible Disc is
automatically powered down after the microprogram is
read, and remains off until a microprogram reload is
required or a diagnostic routine is requested. At this
time, the SCU Microprogram is in a "wait loop," until
commands selecting the SCU are received via the users
channel. The commands are received by the Channel
Interface portion of the SCU, and passed on to the
microprogram for interpretation.

--

STORES MICROPROGRAM

CONTROLS SUBSYSTEM OPERATION

o INTERPRETS AND EXECUTES COMMANDS FROM THE USING SYSTEM

-.-

CORRECTS ERRORS ON HA. COUNT. AND KEY FIELDS

GENERATES ERROR CORRECTION SYNDROME FOR ERRORS IN DATA FIELDS

TRACKS DRIVE STATUS

PERFORMS ARITHMETIC FUNCTIONS

CONTROLLER INTERFACE

SELECTS DRIVE DESIGNATED BY MICROPROCESSOR

COMMUNICATES WITH DISC DRIVE MODULES THROUGH THEIR CON


TROLLERS (ORDERS READ/WRITE DATA AND STATUS)

TO
DRIVES

j
.

FE INTERFACE
PROVIDES INTERFACE BETWEEN FE PANEL AND OTHER SECTIONS OF THE
SUBSYSTEM.

FE PANEL

Figure 6-1 3672 SCU Block Diagram

6-1

6.2 CHANNEL INTERFACE DESCRIPTION

time. Priority is determined by CP Register bits.


When these bits are set, no new selection can
occur.

6.2.1 General
The SCU Channel Interface is the asynchronous link
between the IBM Channel and the SCU Microprocessor.
The Channel Interface hardware consists of Bus In lines
and registers, Bus Out lines and registers, address
compare logic, select logic, priority logic, tag lines and
registers, data transfer logic, control registers, and the
multiplexers necessary to handle data to and from
multiple channels. Refer to the block diagram in Figure
6-2.
NOTE
Refer to Section 3 of this manual for
descriptions and definitions of interface lines, commands, and selection
sequence.

Prevents any channel from having two


successive selection sequences when another
channel is attempting to select the SCU.

6.2.2.5 REGISTERS
The sequencing of the Channel Interface lines in the SCU
to communicate with the channel is controlled by six
special purpose registers.

1. Channel Interrupt (CI) Register


2.
3.
4.
5.
6.

6.2.2.1 ADDRESS COMPARE LOGIC

6.2.2.2 SELECT LOGIC


The select logic enables the initial selection sequence to
continue by setting bit 2 of the CHF Register via INITIAL
SELECT. In addition, INT Register bit 5 (paragraph
6.3.4.2) is set, which allows the microprogram to
recognize that the SCU is being selected.

Channel Flags (CHF) Register


Channel Control (CHC) Register
Channel Tags (CT) Register
Channel Priority (CP) Register
Transmit Request (TR) Register

CHF Register Bits 3-7; Out Tags. Set by the respective


Out tags from each channel. Of these, Data Out/Service
Out (CHF 6-7) are used by the microprogram for
microprogram-controlled data transfer when Channel
Transfer (CHC 0) is reset. For buffered channel data
transfer (CHC 0 set), these bits are used by the interface
hardware to sequence the data transfer.
Channel Control (CHC) Register
The Channel Control (CHC) Register is a special-purpose
register with its bits defined as control functions. Loaded
from 0 Bus in the microprocessor, it controls functions of
Channel Transfer, Channel Write, Lock Channel Switch,
Reset Bus Out Parity Error, Control Unit Busy, Suppressible Data Transfer, and a two-bit Channel 10 Code.

CHF REGISTER BIT ASSIGNMENT

BUS
OUT
PARITY
ERROR

CU
END
OWED

IN
ITIAl
SELECT

ADDR
OUT

CMD
OUT

CHAIN
ING

DATA
OUT

SERVICE
OUT

CHNl
TRANSFER

CHNL
WRITE

RESET
BUS OUT
PARITY
CHECK

CON
TROL
UNIT
BUSY

LOCK
CHNl
SW

Sup
PRESSSIBLE
DATA
TRANS-

ID
CODE
BIT2

ID
CODE
BIT 1

FOR CHNL
MUX

Channel Interrupt (CI) Register.

CI REGISTER BIT ASSIGNMENTS

SCU
ADDR

SCU
AD DR
4

SCU
ADDR
2

SCU
ADDR
1

CU
BUSY
IN
PROG

TAG

ID
CODE
BIT 2

ID
CODE
BIT 1

FOR
SELECTING
CHANNEL

The priority logic is used with a multi-channel SCU. This


logic serves two basic purposes.
Allows only one channel to select the SCU at
anyone time, and serves as a tie breaker if two
channels attempt to select the SCU at the same

Channel Flags (CHF) Register


The Channel Flags (CHF) Register is an assembly of bits
used as branch conditions for the microprogram, gated to
the BR Register by Enable CHF. This register contains
indications for: Bus Out Parity Error, CU End Owed,
Initial Select, Address Out, Command Out, Chaining,
Data Out, and Service Out.

CHC Register Bit 0; Channel Transfer. Set by the


microprogram to initiate buffered data transfer.

The CI Register can be multiplexed onto the A bus.

FOR
SELECTING
CHANNEL

CI Register Bits 6-7; Channel Identification Code. This


is the two-bit output of the channel scanner and tells the
microprogram which channel has been hardwareselected when the Multiple Channel Switch feature is
installed.

CHF Register Bit 2; Initital Select. Set when the


interface has started the selection sequence and reset
when Hold Out falls. This bit is also active for the
duration of a short/busy sequence.

CHC REGISTER BIT ASSIGNMENT

These registers provide the SCU with the capability of


testing channel outbound tags, setting or resetting
channel inbound tags, and controlling the special interface sequences.

6.2.2.3 PRIORITY LOGIC

CI Register Bit 5; Tag. This bit is set when the


TAG/UNTAG switch is in the TAG position.

6.2.2.4 MULTIPLEXER/DEMULTIPLEXER CIRCUITS


These circuits either take mUltiple signal inputs and gate
them onto a single bus with appropriate timing, or they
gate a single bus input into multiple outputs.

6.2.2 logic Description

The address compare circuit compares the hardwired


(jumpered) address of the SCU with the address sent
from the channel, via the Bus Out lines and identified by
the Address Out tag. If the addresses compare equal, the
selection logic will be enabled and Select Out will not be
propagated to the next control unit.

CI Register Bit 4; CU Busy in Progress. This bit is set


when at least one channel is in the process of receiving a
Short Control Unit Busy sequence. The bit is reset when
there is no Short Control Unit Busy sequence in
progress.

CI Register Bits 0-3; SCU Address. This is the SCU


address corresponding to the selecting channel. These
bits are hardwired by the FE at installation time to the
control unit address desired by the customer.

When the Multiple Channel Switch feature is installed


there will be one CHF Register for each channel. In these
cases, the CHF Registers are multiplexed by bits 6 and 7
of the CI Register.
CHF Register Bit 0; Bus Out Parity Error. Set by
Channel Interface parity checking logic if incorrect parity
is transmitted with any byte of data across the channel
with CHC 2 reset.
CHF Register Bit 1; Control Unit End Owed. Set by
Channel Interface hardware, indicating that a "short SCU
busy" has been returned to the respective channel. This
bit is reset when the channel accepts Control Unit End
status.

CHC Register Bit 1; Channel Write. Set by the


microprogram to indicate a Write operation to the
buffered data transfer hardware; reset to indicate a Read
operation.
CHC Register 2; Reset Bus Out Parity Check. Set by
the microprogram to inhibit checking the Bus Out lines
for correct parity. If bit 0 of CHF (Bus Out Parity Error) is
set, setting this bit will reset the error indication.
CHC Register Bit 3; Control Unit Busy. Set by the
microprogram to indicate to the interface hardware that
the control unit is busy. Causes a "short SCU busy" to be
returned to all channels requesting service.

6-2

o BUS

---

CHNL A BO 0-7. P
CHNL B BO 0-7. P
CHNL C BO 0-7_ P

CI500
SEL BO 0-7. P

BUS
MULTIPLEXER

CHNL 0 BO 0-7_ P

CM501

LHNLABUSOUTO-3

CHANNEL IDENTIFICATION CODE

CHNLOBU~~::::

VALID ADDRESS

SELECT
LOGIC
(X4)

ADDRESS
COMPARE
(X4)

----

BUFFER A

CJ500

BUFFER B
CJ500

(X4)

CX502
,~41

CHANNEL
PRIORITY
SCANNER

SELECT OUT (X4)

REQUEST IN (X4)

CHANNEL TAGS OUT IX41

...

CHANNEL
IDENTIFICATION
CODE

PARITv

CP REG

RF. RA " BO

BUS 4
~

ERROR
IX41

.....
~

CHF
REGISTERS
(4)

CI501

..

..,..
~

CX503
CX501

-'"
~

cu BUSY

~NNErr

DATA
TRANSFER
LOGIC

tt

CI502

CX501

i1SERVICE
IN/DATA IN

DISCONNECT IN
FT BUS

lAG
MIJ'. TiPLEXER

LOCK
CHANNEL
CH

CI501

SWr
CHANNEL IDENTIFICATION CODE

CHC
REGISTER

I~:~~~:~R

CI501 10 CODE

OP IN' LOCK CHANNEL SWITCH

CI501

CH A BUS 0-7

MUX
CI501

CT MUX

o BUS

2-7

CI500

CT A-D
0-7

CT MUX

"""\-CX500

CT
LTCH

..

--

CT 0-7 __

EX501

EX501

-'"

TR 07 __

CI500

DATA IN

TR
REG
EX501

MULTIPLEXER

I-

CT REG
(X4)

CT
A-D
0.1

l=0\>
SERVICE IN

BI REGISTER
B

~~

CX500

CI500

'~,"Hj=f}

RD-Bl

CHANNEL
0-3: CM5()( ~E
4: EX501
6,7 CI502

~ "":'''''

I-

OP IN
LOCK CHANNEL SWITCH

CJ501 ' 1 1

DECREMENT CC

+
P

CHNL BUFFER
ROV

PRIORITY
ENABLE

CJ500

1
JK

LOB
25-30

CX500
INTEC

GEN RST
(X4)

BO - CC 24-31

MUX

CI
REGISTER

-..

BUS

o BUS 0-7.

CBR

CHNL SEL

OUT

PROP SELECT OUT SELECT IN ,X4,

INITIAL SELECT

BO
REGISTER
CJ500

CHC

r-CX500

CC 16-23

......

PRIORITY REQUEST

JUMP 0-3.

CHANNEL ENABLE

..

ADOR. GOOD (X4)

CHNL 8 BUS OUTO3 "

CHNL C BUS OUTO-3

JUMP 0-3

JUMP 0-3

CC
REG

D::9

BUS IN
REG 0-7. P

,
BUS IN A 0-7_ P
CHNI.. A

DEMULTIPLEXER

BUS IN B 0-7_ P
CHNL 8

BUS IN C 0-7. P
~

:H".

BUS IN 00-7. P
CHNL S

CX502

CHC 0

Figure 6-2. Channel Interface Block Diagram

6-3

CHC Register Bit 4; Lock Channel Switch. Set by the


microprogram to disable the channel priority scanner to
inhibit selection by channels other than the one already
being serviced. Causes a "short SCU busy" to be
returned to all other channels.

microprogram finishes its operation on the channel


before the channel is disabled.
CT Register Bit 3; Operational In (Controlled by
Microprogram). Reset by interface hardware upon
detecting a channel interface disconnect sequence.

CHC Reg. Bit 5; Suppressible Data Transfer.


CHC Register Bits 6.7; Channel Identification Code.
Set by microprogram or channel interface hardware to
select one of four pairs of CHF and CT Registers when
the Muitiple Channel Switch feature is installed. If either
Operational In (CT 3) or Lock Channel Switch (CHC 4) is
set. the Channel 10 (CI 6,7) is gated into CHC 6,7). If
neither Operational In nor Lock Channel Switch is set,
the microprogram has control of CHC 6,7.

RE
AllOW OPERA
RE
TIONAl
QUEST OUEST
DIS
IN
IN
ABLE
IN
OF
OUEUE STACK
THIS
STATUS
ED
STATUS CHNl

Transmit Request (TR) Register


CT Register Bit 5; Status In. Set by the microprogram to
notify the channel that the SCU status is on the Data In
bus.
CT Register Bit 6; Data In. Set by the microprogram to
notify the channel that data is present on the Data In
bus.

CT REGISTER BIT ASSIGNMENTS

CT Register Bit 4; Address In. Set by the microprogram


to notify the channel that the SCU address is on the Data
In bus.

ADDR
IN

STATUS
IN

DATA
IN

SER
VICE
IN

CP Register Bit 4-7; Priority Windows. Channels AD.When set by multichannel processor, blocks
microprogram detection of INT bits (5) and CHF 2. If the
channel attempts selection with this bit set, Select Out is
blocked and no further channel action takes place until
either: 1) CU Busy (CHC 3) and/or Lock Channel Switch
(CHC 4) is set, in which case a short busy sequence will
begin; 2) Priority Window (CT 7) is reset, in which case
selection may proceed normally; or if the control unit is
already selected, a short busy sequence will begin.

CT Register Bit 7; Service In. Set by the microprogram


to notify the channel that data is present on the Data In
bus.

The TR register can be loaded from the 0 Bus and


multiplexed onto the A Bus.

TR REGISTER BIT ASSIGNMENTS


0

RE
RE
RE
RE
RE
RE
RE
QUEST
QUEST QUEST QUEST QUEST QUEST QUEST
IN
IN
IN
IN
IN
IN
IN
QUEUED QUEUED QUEUED QUEUED STACK STACK STACK
STATUS STATUS STATUS STATUS
ED
ED
ED
CH 0
CH B
CH A
STATUS STATUS STATUS
CH C
CH B
CH 0
CH C

7
RE
QUEST
IN
STACK
ED
STATUS
CH A

channel hardware to the Bus In lines of the channel.


When CHC bit 0 (Channel Transfer) is set without bit 1
(Channel Write), BI interfaces directly YVith the two-byte
buffer in the Channel Interface hardware, which then
interfaces with the Bus In lines on the channel
hardware. When Channel Transfer is reset, BI interfaces
directly with the Bus In lines. BI is loaded by the
microprogram via the 0 Bus. While Channel Transfer is
set, BI is loaded at least one instruction after the
microprogram has verified that the CBR condition is set.
While Channel Transfer is reset, BI is loaded at least one.
instruction prior to the associated inbound tag being set
in the CT Register.
Bus Out (BO) Register
The BO Register is the interface with the Microprocessor
and the Bus Out lines on the channel. This register is the
data communication link between the channel and the
Microprocessor. The register is made up of three
registers: two used as buffers which are loaded
asynchronously by channel tags, and a third which is
synchronous with the Microprocessor. The input to the
synchronous register is multiplexed and controlled by the
processor decode of RA = BO.

Channel Priority (CP) Register


Channel Tag (CT) Register
The CT Register can be loaded by specifying CT in the 0
field or C field (in Format 0) of a microinstruction. When
the Multiple Channel Switch feature is installed there
will be one CT Register for each channel. In these cases,
one CT Register is selected by bits 6 and 7 of the CHC
Register.
CT Register Bit 0; Request In. Queued Status. Set by
the microprogram to raise Request In on the channel
interface. This bit is ORed with the appropriate bit in the
TR Register to raise Request In on the channel interface.
CT Register Bit 1; Request In. Stacked Status. Set by
the microprogram to raise Request In on the channel
interface if Suppress Out is down. This bit is ORed with
Request In, Queued Status (CT 0), and the appropriate
bits in the TR Register to return Request In to the
selected channel.
CT Register Bit 2; Allow Disable. Set by the
microprogram to allow the Channel Disable switch to
affect the Channel Disable latch. This ensures that the

The CP Register can be loaded from the 0 Bus and


multiplexed onto the A Bus.

CP REGISTER BIT ASSIGNMENTS

CHAN
NEl
ERROR
POINT
ER

OVER
RUN
ON
CHAN
NEl
WRITE

PRIO

PRIOR

PRIOR

PRIOR

lTV

lTV

lTV

lTV

WIN
DOWS
CH 0

WIN
DOWS
CH C

WIN
DOWS
CH B

WIN
DOWS
CH A

TR Register Bits 0-3; Request In Queued Status. Set


by the microprogram to raise Request In on the desired
channel interface. Any combination of channels may
have Request In active concurrently.
TR Register Bits 4-7; Request In Stacked Status.
Identical to TR Register bits 0-3 except that the channel
may suppress the request by raiSing Suppress Out.
TR Register Bits 0-7. When displayed on A Bus,
represent the state of the channel tag, which may have
been set by the CT Register. Either the CT or TR Register
may reset Request In regardless of which register
activated this tag.

CP Register Bit 0-1; Unused. Set to zero.


Bus In (BI) Register
CP Register Bit 2; Channel Error Pointer. Indicates
channel-detected errors on Channels C or D.
CP Register Bit 3; Overrun on Channel Write. Indicates
channel freeze under hardware data transfer. This bit is a
flag only. It cannot be set or reset by a 0 Bus statement
to this register.

The BI Register is the interface with the Microprocessor


and the Bus In lines on the channel. Normally the
register is loaded from the 0 Bus as if it were an eight-bit
register, into BI Register A only. In the buffered data
transfer mode, however, the 0 Bus lines are loaded into
both BI Registers A and B to permit buffered operation.
These registers are multiplexed and transferred by the

When in the nondata transfer mode, information is


passed directly to the multiplexer and is latched in the
synchronous register. The other two registers are inactive at this time. When CHC bit 0 (Channel Transfer) and
bit 1 (Channel Write) are set, the Bus Out lines interface
directly with the two-byte buffer in the .channel
hardware, which then interface with the BO entry on the
A Bus in the Microprocessor. While Channel Transfer is
reset, the Bus Out lines interface directly with the BO
entry to the A Bus. During buffered data transfers for
write operations, the data in BO may be transferred in
the same microinstruction in which the CBR condition
was tested.
Channel Byte Count (CC) Register
The' CC Register is a sixteen-bit preloadable counter
which controls the number of bytes transferred during
hardware data transfers with the channel. While CHC bit
o (Channel Transfer) is set, the CC Register is
decremented each time the channel responds to Data In
or Service In with Data Out or Service Out, respectively.
Since the decrementing of the CC Register (when
Channel Transfer is set) is asynchronous to the
Microprocessor cycle, the CC Register can be loaded to or
from R23 only when the Channel Transfer bit is reset.

6-4

6.2.3 Data Transfer Organization


6.2.3.1 GENERAL
The SCU Channel Interface provides the capability for
buffered data transfer between the user's system
channel and the SCU Microprocessor. Buffered data
transfers to the channel are initiated through the CHC
and CC Registers.
Data transfers between the channel and SCU fall into
two categories:

Register. The CC = 0 latch, which is INT Register bit 0, is


synchronous with the Microprocessor cycle so that the
microprogram may branch on the condition at any time.
The interface between the channel busses and the
Microprocessor are the two registers, Bus In (BI) and Bus
Out (BO). While the Channel Transfer bit in CHC is reset,
these two registers directly reflect the Bus In and Bus
Out lines of the channel. When Channel Transfer is set
in CHC for Read operations, BI Register interfaces with
the Bus In lines on the channel via the two-byte buffer.
For Write Operations, BO Register interfaces with the
Bus Out lines on the channel via the two-byte buffer.

Read Operations-Cause data to be transferred


from the SCU to the channel via the Bus In
lines.
Write Operations-Cause data to be transferred
from the channel to the SCU via the Bus Out
lines.

The microprogram initiates buffered data transfers by


setting Channel Transfer bit 0 in the CHC Register. At the
same time, the microprogram establishes whether the
transfer will be a Write or a Read by respectively setting
or resetting Channel Write bit 1 in CHC.
Once the microprogram has initiated a buffered data
transfer. the channel hardware sequences Service In and
Data In to effect the transfer of data between the onebyte channel bus and the two-byte buffer within the
Channel Interface hardware. Each time a byte is
transferred, the CC Register is decremented to show how
many bytes are remaining to be transferred between the
channel and the SCU. The decrement condition is
logically Service Out AND Service In, or Data Out AND
Data In. When the CC Register is decremented to zero, a
latch (CC = 0) is set in the INT Register of the
Microprocessor, and the Channel Interface hardware is
inhibited from raising Service In or Data In, thus stopping
the transfer of data. For example, if a count of five is
placed in the CC Register, five bytes will be transferred
between the SCU and channel on both Read and Write
operations. The decrementing of CC is asynchronous to
the Microprocessor; therefore, the Channel Transfer bit
in CHC must be reset to transfer CC to or from R23.
The set condition for the CC = 0 latch is logically Service
In AND Service Out, or Data In AND Data Out, when the
value of the channel counter is equal to one. The CC = 0
latch is reset when the microprogram reloads the CC

The detailed operation of the data transfer hardware for


Read and Write operations is described below.

To initiate buffered data transfers from the SCU to the


channel, the microprogram loads the CC Register with
the number of bytes to be transferred and sets the
Channel Transfer in the CHC Register. Loading CC
causes the CC = 0 latch to be reset if it was on, and
setting Channel Transfer gates the two-byte BI buffer
into the data transfer path. After the correct count has
been loaded in CC and Channel Transfer has been set,
the Channel Interface hardware sets the Channel Buffer
Ready (CBR) and Queue Empty (QE) branch conditions in
the Microprocessor. The CBR conditions indicate that one
or both buffers have not been loaded by the
microprogram, and the QE condition indicates that both
buffers are empty.
When the microprogram responds to the CBR condition
by loading a byte into BI, the Channel Interface hardware
places this byte on Bus In to the channel and raises Data
In. The microprogram loading BI causes the QE indication
to be reset since one buffer now contains a byte of data.

The channel issues a Halt I/O or Halt Device


instruction to the SCU by raising Address Out.
in which case the microprogram will reset the
Channel Transfer bit in CHC

BUS IN

_ - Q - E - - - - - I TRANS FE R I-----S-D::-;
__~-I..,.~N=-E""IN--RD=BI
H~O:6:~~EI---------~~

CC REG

6.2.3.2 READ OPERATIONS


A simplified block diagram of the Channel Interface
circuits involved during a Read operation is shown in
Figure 6-3. During Read operations, the Channel Interface hardware accepts data from the microprogram via
the BI Register and transfers this data asynchronously to
the channel over the Bus In lines. The BI Register is
structured as two eight-bit buffers between the 0 Bus
and the Bus In lines. Since the data from the disc drive is
processed regularly by the microprogram (average of 1
byte every 1 .24 microseconds). these two buffers allow
the time between byte transfers to the channel to vary
considerably.

The channel truncated the data transfer by


raising Command Out in response to Service In
or Data In, in which case the CC Register is not
decremented, the QE condition is not set, and
the CBR condition can be either on or off. When
the microprogram detects truncation through
INT bit 5 (Channel Interrupt) or CHF bit 4
(Command Out). It will reset the Channel
Transfer bit in CHC, ending the buffered data
transfer.

D BUS

CBR

FT BUS

Figure 6-3. Read Data Transfer Block Diagram

CBR, however, remains set since there is still one


available buffer for more data. Loading BI initiates the
buffered data transfer.
At this point. the channel may respond to Data In with
Data Out (or Command Out indicating truncation) indicating the byte has been accepted (or rejected in the
case of Command Out) by the channel, and thereby
cause the QE condition to be set again. Alternately, the
microprogram. may load another byte into BI (since the
CBR condition is still set) and thereby cause the CBR
condition to be reset. If the former condition occurs, the
CC Register is decremented. If decremented to zero, Data
In and Service In are inhibited from being set again until
the CC Register is reloaded. On the other hand, if the
latter condition occurs, both buffers will contain data
which had not been accepted by the channel, and the
microprogram may not load another byte of data into BI
until the channel accepts at least one of the buffered
bytes.
The data transfer to the channel continues with the
Channel Interface hardware alternately raising Data In
and Service In to transfer successive bytes until one of
the following conditions occurs:

The CC Register is decremented to zero, in


which case the data transfer has been completed.

For Sense commands, where the data transfer is a Read


operation over the channel and the data need not be
transferred regularly, the microprogram indicates suppressible data transfer to the channel hardware by
'setting bit 5 in CHC. The transfer operates identically to
that described above, except that the Channel Interface
hardware is inhibited from setting Data In or Service In to
transfer a new byte while Suppress Out is set at the
channel. When Suppress Out is off, however, the data
transfer hardware continues unaffected.

6.2.3.3 WRITE OPERATIONS


A simplified block diagram of the Channel Interface
circuits involved during a Write operation is shown in
Figure 6-4. During Write operations, the Channel Interface hardware accepts data asynchronously from the
channel over the Bus Out lines and transfers this data to
the microprogram via the BO Register. The channel
hardware contains a two-byte buffer between the Bus
Out lines and the BO Register. Since data is required for
writing by the controller and disc drive regularly (average
of 1 byte every 1 .24 microseconds), these two buffers
allow the time between byte transfers from the channel
to vary considerably.
To initiate buffered data transfers from the channel to
the SCU, the microprogram loads the CC Register with
the number of bytes to be transferred and sets Channel
Transfer and Channel Write in the CHC Register. Loading
CC causes the CC = 0 condition to be reset if it was set
prior to the loading. Setting the Channel Transfer and
Channel Write bits in CHC gates the two-byte BO buffer
into the data transfer path. CBR is reset by the hardware
one instruction cycle after Channel Transfer and Channel

6-5

1"'"-''--

BUFFER
A

MUX

BUS OUT

DATA IN
--"":S:':"E:":'R':":V"":'IC:":'E-I-N--t

A BUS

---r--

CBR

TC~~~:~~ t -_ _ _ _ _Q;::;E::....-_-tl~

--~~~----tHARDWARE~

CC REG

_ _ _ _ _R_A_=_BO_ _
FT BUS

Write are set. When the microprogram sets the Channel


Transfer and Channel Write bits with the CC = 0
condition reset, the channel hardware raises Data In,
requesting the first byte of data from the channel. Until
the first byte of data is transferred by the channel, the
CBR condition remains reset, indicating to the
microprogram that data is not yet available from the
channel. As soon as the channel transfers the first byte
of data, the Channel Interface hardware gates the byte to
the BO Register. The CBR condition is then set to
indicate to the microprogram that data is available in the
BO Register from the channel. Also, since there is still
one buffer available for data, the channel hardware
immediately raises Service In to request another byte of
data to be loaded into the second buffer.

The channel truncates the data transfer by


raising Command Out in response to Service In
or Data In, in which case the CC Register is not
decremented and the CBR condition is not set if
it was reset. The microprogram detects truncation through INT bit 5 or CHF bit 4.

The channel overran the data transfer rate of


the disc drive by not responding to Service In or
Data In with Service Out or Data Out before a
byte of data was required. When the
microprogram detects overrun, it will execute a
Channel Freeze control statement and set CP
Register bit 3 (Overrun on Channel Write).'
Executing this control statement causes the
present Service In or Data In transfer sequence
to be completed, and inhibits the initiation of
any new transfer. sequences. If Service In and
Data In are both reset when the Channel Freeze
control is executed, data transfer is immediately
halted. After the Channel Freeze control is
executed on data overruns, the Queue Empty
condition is set when the present transfer
sequence has been completed and no new
sequences are to be initiated. When QE is set,
the microprogram will then reset Channel
Transfer and Channel Write.

At this point the microprogram may respond to the CBR


condition by gating the BO Register to the A Bus. thereby
causing the CBR condition to be reset. Alternately, the
channel could transfer another byte to the SCU (since
one buffer is still available for new data), thereby causing
the CBR condition to remain set. In either case, the CC
Register is decremented with each byte transferred by
the channel, and if decremented to zero, Data In and
Service In are inhibited from being set again until the CC
Register is reloaded. If the latter case occurs, however,
both buffers will contain data which had not been
accepted by the microprogram; consequently, the
Channel Interface hardware cannot request another byte
from the channel until the microprogram has accepted at
least one of the buffered bytes.

The data transfer from the channel continues with the


Channel Interface hardware alternately raising Data In
and Service In to request successive bytes until one of
the following conditions occurs:

Figure 6-4. Write Data Transfer Block Diagram

pleted. When the microprogram detects this


through INT bit 0, it resets the Channel Transfer
and Channel Write bits in CHC.

The CC Register is decremented to zero, in


which case the data transfer has been com-

The channel issues a Halt I/O or Halt Device


instruction to the SCU by raising Address Out,
in which case the microprogram will reset the
Channel Transfer and Channel Write bits in
CHC.

For Seek commands where the data transfer is a Write


operation over the channel and the data need not be

transferred regularly, the microprogram indicates Suppressible Data transfer to the channel hardware by
setting bit 5 in CHC. The transfer operates identically to
that described above, except that the Channel Interface
hardware is inhibited from setting Data In or Service In to
transfer a new byte while Suppress Out is set at the
channel. When Suppress Out is off, however, the data
transfer hardware continues unaffected.

6.2.4 Channel Interface Branch Conditions


6.2.4.1 CHANNEL BUFFER' READY (CBR)

Channel Read
Indicates that one or both channel buffers are empty.

Channel Write
Indicates that one or both channel buffers are full. The
buffer may be unloaded in the same instruction as the -B
(CBR) statement.
6.2.4.2 QUEUE EMPTY
Indicates that both channel buffers are empty during a
Read operation or the CC Register contents are zero.
6.2.4.3 CHANNEL INTERRUPT (INT 5)
Channel Interrupt is active under one of the following
conditions:
1. Initial Selection AND Not Short Busy Sequence
AND Not Operational In.
2. Operational In AND Halt I/O + Truncation.

6-6

6.3 MICROPROCESSOR DESCRIPTION

Buffer Storage-Provides temporary storage for


Drive status. a working area for some types of
error correction. and information related to drive
orders.

Flexible Disc Interface-Provides control and


data interface for the Flexible Disc Unit.

6.3.1 Microprocessor General Description


The Microprocessor controls overall operation of the
3672-based system via the microprogram Writable Control Storage (WCS). The Microprocessor contains the
registers. data busses. and storage necessary to interpret
and execute subsystem commands. Commands are executed by translating the microinstruction fields into
hardware control signals which perform the actions
required to execute microinstructions.
The. Microprocessor is organized
functional areas (Figure 6-5):

into the following

Writable Control Storage-Provides storage for


the microprogram and diagnostic routines.

Data Paths-Composed of the ALU. registers.


and
busses
necessary to
operate
the
Microprocessor.

WRITABLE
CONTROL
STORAGE

---

J~

FLEXIBLE
DISC
INTERFACE

"
DATA
PATHS

...

,~

,r

INPUT TO
B MUX

INPUT TO
F MUX

INPUT TO
B MUX

INPUT 10
F MUX

BRANCH ADRS
TO ADRS REG

MICROPROCESSOR
CONTROL FIELDS

TRANSLATION

6.3.2 Microinstructions
6.3.2.1 DESCRIPTIONS AND FORMATS
The microprogram controls the SCU hardware through
the sequencing of microinstructions. These microinstructions are divided down into fields whereby each field
controls a specific function in the hardware. In the SCU
the data for these microinstructions is stored in WCS.
There are four microinstruction word formats. Each
format, and the fields contained therein. raises specific
controls which operate upon the logic circuits.

MICROINSTRUCTION
FIELDS
BRANCH ADRS
TO ADRS REG

MICROPROCESSOR
CONTROL FIELDS

TRANSLATION

Because of the large number of controls needed within


the SCU, the 34 usable bits of the microinstruction word
are translated into 57 usable bits. This translation
concept is used to allow a smaller memory word size.

MICROINSTRUCTION
FIELDS

Specific details about the four microinstruction formats


and the interpretation of each field are contained in
paragraph 4.1 of Section 4. For convenience in the
discussion of microinstructions in this section, the four
microinstruction formats and their expansion from a 34bit word stored in WCS to a 57-bit word used for
hardware control, Figure 4-3 is repeated in this section
as Figure 6-6.

MICROPROCESSOR
CONTROL FIELDS

o
INPUT TO
B MUX

EQUAL TO HIGH ORDER 6


BITS OF ADRS REG

BRANCH ADRS
TO "'DRS REG

INPUT 10
F MUX

TRANSLATION

MICROINSTRUCTION
FIELDS
BRANCH ADRS
TO ADRS REG

INPUT TO
F MUX

6.3.2.2 MICROINSTRUCTION FIELD TRANSLATION


AND BRANCHING
MICROPROCESSOR
CONTROL FIELDS

Translation

RA

FORMAT 3
RD

RK

RX

RT

RF

RBC

RN

FORCES NO DEST FOR TBUS

BUFFER
STORAGE

Figure 6-5. Microprocessor Block Diagram

The microinstruction translation hardware translates the


34-bit microinstruction word into 57 hardware control
lines. This translation occurs after the microinstruction
has passed through the ECC circuits and the data bits are
latched at the output of the ECC circuits (Figure 6-10).
The translation during each machine cycle is controlled
by the microinstruction format code. The 57. hardware
control bits are gated through the translation matrix by

TRANSLATION

MICROINSTRUCTION
FIELDS

Figure 6-6. Microinstruction Expansion

6-7

the format code as shown in Figure 6-7. The outputs of


the translation matrix control the Microprocessor
hardware while performing ALU operations. register-toregister transfers. and so forth.

r---------,
B FLO

Branching

The first stage of multiplexing is accomplished by four


circuits: the BRA multiplexer. the BR mUltiplexer. and
gates A and B. All four circuits are enabled by PS 4-7.
The BRA multiplexer gates inputs from RO. R1. R2. R3.
G4. G7. 0, and eight interrupt conditions via the four PS
bits through the PSA latch. The BR multiplexer gates the
output of the BRA multiplexer and 01 Register via
ENABLE 01. This enable is also developed from PS 4-7
through an encoder. Gates A and B enable the output
from the BR multiplexer or CHF via ENABLE CHF. This
enable is also developed from PS 4-7. The table in
paragraph 6.3.4.1 gives a complete listing of the encoded
combinations of PS 4-7 required to gate each input to the
multiplexers.
The result of this first stage of multiplexing is called BR +
CHF. The BR + CHF lines are multiplexed with the branch
conditions. or the INT Register. by the BC field to
determine the branch condition specified in the
Microinstruction word. If the branch condition is met. the
BRANCH ADDRESS SET signal is generated to load the
specified branch address.

B FLD

[ )

RN FLO 05

RN FLO 68

FMT 23

RN FLO 911

FMT 3

RX FlO

FMTO

RK FLD

FMT 0 + 1 + 2

RBC FLO

FMT1+2+3

RA FLO

FMT 0

FMT 3

The branch multiplexing circuits are shown in Figure 6-8.


These circuits multiplex inputs from several different
sources under control of the Page Select (PS) Register.
This result is multiplexed with a branch condition or the
Interrupt (INT) Register contents. under control of the
microinstruction BC field. to produce a final branch
address.

BA FLO
FMT 1 + 2 + 3
BA2 + BA3 FLD

X FLD

ECC
HOWR

Z FLO

Be FLO

A FLD
1+3

FMT 0

)RT/RF
XLTR

SEP RT
AND RF
DECODES

I
I

Figure 6-7. Microinstruction Translation

Branch on 0 Bus (0 BUS)


The low-order 8 address bits to WCS are
determined by the information resulting from an
ALU operation logically "ORed" with the Branch
Address field of WCS. The remaining bits are
determined in the normal way, depending on
which format is being executed.

)L____________~~:.J
I

FMT2

6.3.3.2 READING FROM WCS

The read access of WCS words takes place whenever a


valid address exists at the input of the memory board.
This address is provided from the RAM Address Register
which may be conditioned by the following:

r---------------,
I
I
I

The Writable Control Storage (WCS) is a volatile


semicondu'ctor storage system organized on seven
printed circuit boards. Each of the seven boards contains
ninety-six 256 x 1 RAM chips. The chips are arranged on
each board in a 6 x 16 matrix so that each board stores 6
bits of 4096 words. The seven boards together provide
4.096 words of 42 bits. The 42-bit word is subdivided to
provide 34 data bits (normal microinstruction or
microdiagnostic word as it resides in storage) plus 8 bits
used for error correction. A WCS extension feature is
available to increase the size of WCS from the nominal
4.096 words.

Refer to the WCS block diagram in Figure 6-10.


.:..._~

CC FLO

Execution of this microinstruction is carried out in the


following sequence. First. the microinstruction is read
from WCS location ODE1. Next. the ALU operation
specified in the F field is performed on the registers
designated by the A and B fields; namely. the contents of
the BO Register are added to those of the GI Register and

L.. _______RT501

6.3.2.3 MICROINSTRUCTION EXECUTION


EXAMPLE

6.3.3 Writable Control Storage Description


6.3.3.1 GENERAL

BA3 FLO

C FLO

An example of how microinstructions are executed from


the information contained in their fields is shown in
Figure 6-9 . The microinstruction field information can be
summarized as below.

the result stored in the CHC Register. Then the branch


specified by the BC field is performed. Since the branch
is unconditional. formation of the branch address takes
place immediately in the Address Register. This amounts
to gating the contents of the BA3 and BA fields to form
the high-order and low-order parts of a 12-bit branch
address. (Bits 11-0 of this register run from left to right.
opposite from all other SCU Registers.) Expressed in
hexadecimal form. this branch address is 06CA. which is
where the program jumps to read the next microinstruction in the program.

Initial Microprogram Load (IMPL)


The WCS address during IMPL is determined by
a loader source register. The entire address field

6-8

RD 0-7,P

6.3.3.3 WRITING INTO WCSMEMORY

BRA
MUX

During a Write into the WCS, the address input to the


RAM board is decoded in the same fashion as during the
Read to select the chips in which data is to be written.
The following input relationship is necessary to accomplish a Write function.

Rl0-7,P
R20-7.P
R30-7,P

A RAM word is written when the WCS signal is sent to


the RAM PCB.

G70-7.P
DO-7,P
INT COND (B)

The data written is the data contained in a 42-bit Write


Data Buffer located in the FD interface. A Write function
takes place in two different ways:

BRA
0-7.P
SEL GC502

1. During IMPL the bootstrap is assembled from


theFD unit and stored into RAM under
hardware control.

PSA 4-7

DIO-7.P

2. When the microinstruction being executed contains the WCS control statement.

BR
MUX
GM502

6.3.3.4 ERROR CORRECTION CODE (ECC)

ENABLE 01

ENABLE CHF

BR CHF 0-7

BRANCH
ADDR
SET
MUX 1

BRANCH
ADDR
SET
MUX 2

SET TO ADDR REG


BRANCH AD DR

CHF 0-7
RA500

RA500

BC FIELD - -....-------~

Figure 6-8, Branch Multiplexing Circuits


is determined by the contents of this register.
Other address sources are inhibited from
loading the Address Register during this mode.

Start Address (START ADRS)

Subroutine Register Address Source (SR EXT)


With the B(ADDR = SRI statement, the address
is derived externally from this card by means of
the SR Register. All other normal operational
means of address determination are inhibited.

ECC Generation

Branch Address (BRCH)

Once the address is received at the input of the RAM


boards, address bits 8, 9, 10, and 11 are decoded to
provide a select for one column of six RAM chips in the 6
x 16 (per board) memory matrix (Figure 6-11). The other
8 address bits (0-7) are applied at the chip level to select
the data bit stored at one of 256 addresses. This decoded
select and chip address provides 6 bits read-out of each
of the seven memory boards .for a total of 42 bits_ The
output of the RAM boards is set into the RAM Data
Register to be acted upon while the next address is being
conditioned into the Address Register.

The branch address is determined by whether


the list of branch cQnditions shown in Figure 69 are satisfied and which microinstruction format mode is being operated on.

A read access from WCS requires 110 nanoseconds after


the address is presented to the RAM board for data to be
good at the output of the card. For the relationship in
time of addressing, refer to paragraph 6.3.7.

The address incrementer is a parallel load


counter loaded by the present address initiated.
The counter then is incremented by 1 to determine the next address during Format 0 execution or when branch conditions are not satisfied.
During the memory scan mode, the incrementer
is a constant source for addressing.

The Start Address is loaded from the FE panel


control and selection. The logic on this board
accepts loading of the start address mode.

Address Incrementer (INCR)

The microinstruction and microdiagnostic words read


from the Flexible Disc are accompanied by an ECC read
from the Flexible Disc. As a word is passed through the
Flexible Disc Interface, a new ECC pattern is generated
and compared with the ECC pattern read from the
Flexible Disc. A miscompare results in an FD Read Error,
and a Check-1 condition is set if doing an IMPL
(paragraph 6.6). If not doing IMPL, the error causes a
Check-2 condition.

Due to the complexity of ECC generation, it will be


explained by example.
The ECC
Table 6-1

is generated using the matrix shown in

An example data word applied to the ECC matrix is


shown in Table 6-2. Each 1 bit of the data word is applied
to three rows of the matrix. Each 1 bit in a horizontal row
(A through H) is counted and the ECC bit (P 1 through Ps)
for that row is set (or reset) to make the bit count for a
given row even. As an example, bit position 2 of the
example data word is a 1 and is applied to rows A. G and
H of the ECC matrix. Bit position 23 of the example data
word is a 1 and is applied to rows A. C and D of the ECC
matrix, etc.

6-9

ECC LATCHED DATA BIT BUS: 34 BITS FROM WCS

MICRO INSTR BITS

r----------------------------------------J~~--------------------------------------------------------~
10

1~

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

"
FORMAT 3 MICROINSTRUCTION

FIELD

CONTROLS
BINARY VALUE
OF WORD
STATEMENT FOR
INSTR. BLOCK

OP
CODE

BA3

BC

BA

WORD
FORMAT

A BUS

o BUS

B BUS

ALU

BRANCH ADDR
BITS (11-6)

BRANCH
FIELD

BRANCH ADDR
BITS (5-0)

SELECTS BO
REG TO A BUS

SELECTS CHC REG


FOR RESULTS OF
ALU OPERATION

SELECTS
Gl TO
B BUS

HIGH ORDER
SIX BITS OF
RN (11-6)

SELECTS
ALU FUNCT
ADDR

"-

SELECTS BRANCH
CONDITION
/

"-

MICROPROGRAM W ORO FETCHED


FROM WCS ADDRESS OOE1

LOW ORDER
SIX BITS OF
RN (5-0)

~'

HEX
CODE

00

01
02
03
04
05
06
07
08
09
OA
OB
OC
00
OE
OF
10
11
12
13
14
15
16
17
18
19
1A
1B
1C
10
1E
1F

BC CODE
B (BRO)
B (BR1)
B (BR2)
B (BR3)
B (BR4)
B (BR5)
B (BR6)
B (BR7)
B (0 =0)
-B (0 =0)
B (0 =FF)
-B (0 =FF)
B (CARRY)
-B (CARRY)
-B (COMPARE. EO)
B (COMPARE. HI)
B (lNT)
-B (lNT)
B (CNT =0)
-B (CNT =0)
B (LD.SR_*1) LOW
B (DBR)
B (lLACT)
-B (DBR + INT)
B (CBR)
B (0. EMPTY)
B (NO. BRANCH)
B (LD.SR.*+1) HI
B (UNCOND)
B (D_BUS)
B (ADR =SRI

THE BA AND BA3 FIELD IS GATED INTO THE


ADDRESS REGISTER FOR FORMAT 3 IF BC
FIELD CONDITION IS MET.

ADDR
REGISTER

0
1

Di

n-{

TO
INCREMENTER

3
4
ADDRESS

5
6

BUS

TO STORAGE
CONTROL (WCS)

8
9

10
11

TO DISPLAY
AND ADDR ERROR
CIRCUITRY

RA502

Figure 6-9. Microinstruction Execution Example

6-10

Row A of the example has bits 2 and 23 noted, giving an


even count for that row; therefore P1 is a zero. However,
row H has bit position 2, 5, and 6 noted, giving that row
an odd count. Therefore, Ps is turned on to make the
count for row H even.

INCR
BITS 0-11
SR EXT
3120

. -

IPL
BITS 011
BRCH
ADRS 011
START
ADRS 011

...

ADDRESS 10-71

118-11,--

.......
ADDRESS
REGISTER

DEC
RBSOD

CS

WCS
-RAM-

DATA OUT BITS 0-41

...

RBSOD

--

RAM
MUX

RM

A~

DATA
REGISTER

ECC Checking
RDSOD

DATA BITS
233

FD
ECC
GENERATION
AND
VERIFICATION

RA502

...

Other bits of the data word are applied to the matrix in a


similar manner. The generated ECC is shown at the right
of the example .

DATA BIT50-41

SYNDROME
GEN

RD50D

FD

A~

SEP DATA. SEP CLOCK _

ECC Checking From the FE Panel

"

ERROR
CORRECTION
AND
DETECTION
HARDWARE

FD
INTERFACE

After the data from the WCS is set in the Data Register,
the data and its ECC bits are passed through a syndrome
generating circuit. The generated syndrome is applied to
the correction circuits. If the syndrome is all zero's, no
'error exists. If the syndrome has 3 and only 3 bits
present, a correctable error exists and is therefore
corrected. If any combination of syndrome bits are
present, except 3, an uncorrectable error exists and the
SCU will stop.

Pl- PB

t
DATA
AND ECC
FROM
FD

The ECC used with WCS allows detection and correction


of single bit errors (one bit dropped or picked up).
Multiple bit errors can be detected but not corrected.

FC

When necessary, the contents of a WCS word, after error


correction, can be displayed on the FE panel (Display
Roller at position C). The syndrome bits (P1-P8) used in
error correction may also be displayed (Display Roller at
position H).

RESOD

LDB
0-33

Also note that the SCU can be forced to stop on single


(correctable) ECC errors when in FE mode with the CHK2
STOP switch in the on (up) position. This enables the FE
to detect single bit errors during scheduled maintenance.

"
D BUS BIT 07

Figure 6-10. WCS Block Diagram

MICRO
PROCESSOR
CONTROL
HARDWARE

---

MICROINSTRUCTION

TRANSLATION

RT50D
RT501

6.3.3.5 WCS EXPANSION


The WSC expansion feature enlarges the size of WCS by
providing increased storage to support additional
features. These features are implemented through the
use of additional RAM boards. As shown in Figure 6-12,
each RAM board can store 512 42-bit words. These
boards are addressed by ADDR BIT 0-12. Bits 0-7 are
used to select one of 256 words on a board. Bit 8 selects
either the upper or lower 256-word portion of a board.
Bits 9-12 are used to select one of the RAM boards. (The
detailed selection scheme is shown in the logics for the
RX board.)

6-11

CHIP SELECT
COLUMN 1

10

11

12

13

14

15

16

0-7

ADDRESS INPUT
FROM ADDR REG

--+-DOO

ROW A

AD DR BIT 0-12

WCS OUTPUT
JUMPERED IF
NO WCS EXT

wcs
RB

8-11

DO 0-41

CS1, CS2
D

E
CS3

RAM
MUX

MAXIMUM
4

BOAR~L-____~

RM

3
WCS
EXT

DO 041 U

ADOR BIT 13

Figure 6-12. WCS Extension Block Diagram


2K

1K
A

3K
.A

4K
A

B Bus

LOGIC SHOWN CONTAINED ON RB500

-+-o-RIVER

WCS SIGNAL

This bus transfers the RK control bits (specified by the Z


field of the microinstruction) to the B entry of the ALU in
Format 0, 1 and 2 microinstructions. In Format 3
microinstructions, the data from a register selected by
the RB control bits (as specified by the B field of the
microinstruction word) is transferred to the B entry of the
ALU.

WRITE ENABLE
ROW A - F

NOTE: This diagram is typical of the RAM card


addressing. There are seven identical RAM cards.

o
Figure 6-11. RAM Board Column Select

The selected word from the WCS extension is sent to the


RAM data multiplexer. The multiplexer is also fed with
the output from the 4K WCD, end is used to select words
from either the WCS or the WCS extension. Selection is
accomplished via address bit 1 2, which selects WCS if a
zero or WCS extension is a one. If the WCS extension
feature is not present in the.SCU, jumpers are connected
on the multiplexer board back panel to bypass the
multiplexer_

6.3.4 Data Paths

8 bits wide. Figure 6-13 illustrates the functional


arrangement of the buses.
A Bus

6.3.4.1 BUSES
The Microprocessor has five major buses which allow
the transfer of data throughout the SCU. These buses
consist of the A Bus, B Bus, D Bus, T Bus and BR Bus.
The T Bus is 36 bits wide to allow the transfer of 32 data
bits and 4 parity bits. The A, B, D and BR buses are each

Bus

This bus transfers data from the- output of the ALU to a


register determined by the RD control bits (as specified
by the D field of the microinstruction word).

This bus transfers data from a register selected by the RA


control bits (as specified by the A field of the
microinstruction word) to the A entry of the ALU in
Format 0, 1 and 3 microinstructions. In Format 2
microinstructions, the RA field is forced to zero to place
all zeros on the A entry to the ALU.

T Bus
This bus transfers data. from the output of the F
multiplexer to a register determined by the RT control
bits. In Format 0 microinstructions, the X field is placed
on the T bus and the RT control bits are specified by the
C field of the microinstruction. In Format 2 microinstructions, the RT control bits are decoded from the CL field of
the microinstruction to effect the necessary register-toregister transfer.

T BUS 32 BITS

GENERAL
PURPOSE ~
REGISTER S

FROM CHAN
FILE.FO&FE
INTERFACES

"

u
GO

(S BITS)
GEN. PUR.

I"
II
l

o BUS

1"
G1

G2

(S BITS)
GEN. PUR.

II "
~I

G4

G5

G6

"

G7

"

RO

(S BITS)
GEN. PUR.

(8 BITS)
GEN. PUR.

(S BITS)
GEN. PUR.

(S BITS)
GEN. PUR.

(8 BITS)
GEN. PUR.

(8 BITS)
GEN. PUR.

(S x 16 BITS)
SCRATCH

I ..

I"
"I

"

_+ I ~~

JI

G3

A~ ~

"

"

PS
(8 BITS)
MPX R03
& BR MPX

.4~

.0

"

~l

FT BUS 32 BITS

'.

Rl

R2

R3

(S x 16 BITS)
SCRATCH

(Sx16BITS)
SCRATCH

(S x 16 BITS)
SCRATCH

"
I

I
I
"
"
"
I A~

~I

~I

FUNCTION
DECODES FROM ..
MICROPROGRAM ...
WORD

"

A BUS

ALU
A
S BITS

"

D~ ~

I I

II

MICROWORD
CONSTANT
(16 BITS:
OP.Q)
(8 BITS:
OP12)

"

I
~~

"

I .. ,",u" ""

0=0

BRANCH CONDo

D=F F

BRANCH CONDo

A<B

BRANCH CONDo

A=B

BRANCH CONDo

CARR Y

BRANCH CONDo

TO ADDRESS REGISTER

MCM

CCK

BUF

BAR

INT

18 BITS)
MACH CHK
MPX REG

MACH CHK
OR CHC

(512 WORDS
32 BITS
BUFFER)
STORAGE

(9 BITS)
ADD FOR
BUF

(8 BITS)
INTERRUPT
REG.

IMK

CNT

SR

(8 BITS)
(8 BITS)
116 BITS)
INTERRUPT MACH CYCLE SUBRTN RET
ADDRESS
REG. MASK
COUNTER

.l

A~

T BUS 32 BITS

BRANCH
CONDITION
GENERATOR

B BUS S BITS

.n

o BUS S BITS

FROM CHAN
FI LE.FD & FE
INTERFACE

A~

.....

TO ADDRESS REGISTER

1"

"
SPECIAL
PURPOSE
REGISTER S

S BITS

T TT T

I I

II

G2.G3-----....J
INCREMENT BAR CONTROL - - - - - - - - - - '

DTC

= ZERO

..
...

........

SUBROUTINE
REGISTER
TO ADDRESS REGISTER
TO CHANNEL. FILE,
FD. & FE INTERFACES

TO CHANNEL &
FILE INTERFACES

.:

~~T;~~~6T CONDITION)
~N~E~ZERO

INCREMENTER

~~~I~~~~UTINE

.
~

_______________________________________________~~

CNT = 0 BRANCH
CONDITION

INTERRUPT BRANCH
CONDITION

CHAN.INTP.
RESPONSE
CHECK 2

Figure 6-13. Microprocessor Buses and Registers

6-13

BR Bus

TABLE 6-1

This bus transfers data from one of the eight-bit registers


to the branching multiplexer. The register to be multiplexed onto the BR bus is determined by bits 4, 5. 6 and
7 of the PS Register. The registers available to the BR
multiplexer and the associated hexadecimal code required in the PS Register are shown below.

WCS ECC MATRIX

BIT POSITION
ROW 0

2
1

1
1

P P P P P P P P
1 2 3 4
7 s
5 6
30 31 32 33 34 35 36 37 3B 39 40 41

1
1

0
1

1
2
3

REGISTER
GATED TO BR BUS

HEXADECIMAL
CODE IN
PS REGISTER
BITS 4, 5, 6 AND 7

22 23 24 25 26 27 28 29

10 11 12 13 14 15 16 17 IS 19 20 21

6
7

G4
Rl
R2
R3
INT
RO
G7
0
CHF

9
A

B
C

01

E
F

TABLE 6-2. EXAMPLE DATA WORD APPLIED TO WCS ECC MATRIX

6.3.4.2 REGISTERS

1!0

BIT POSITION

0 1 2

EXAMPLE
DATA WORD

1:

10 11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

P
3

P
4

P
5

P
6

P
7

34

35

36

37

3B

39

40

41

0
1

PI

P
2

1
1

1 1

0
0
1

The storing of data and the controlling of interface


hardware in the SCU is accomplished with general
purpose and special purpose registers. The general
purpose registers are not identified with any special
hardware function of the machine. At one point in the
microprogram, a general purpose register may be used to
store the count of time duration; however, at another
point. that same register may be used to store data which
has been obtained from the CPU for a compare operation.
Special purpose registers, however. are identified with
special hardware functions in the machine and, in fact.
may be directly attached to signal lines on an external
interface on the machine. A functional block diagram of
the general purpose registers and special purpose
registers is shown in Figure 6-13.
General Purpose Registers
Register File R03. Register File R03 (RO. R1. R2. and R3)
is a storage array of 16 four-byte words which are paged
by bits 0-3 of the PS Register. The addressed word of the

6-14

register file can be referenced a byte at a time by


separately specifying RO, R1, R2, or R3.

G7 Register. The G7 Register is a one-byte register


which can be accessed through the A or BR Bus and can
be loaded from the 0 Bus.

When RO, R1, R2 or R3 are specified, the corresponding


eight bits are gated to the. A Bus or loaded from the D
Bus. The word operated upon in a microinstruction is
determined by the contents of the PS Register at the end
of the execution of the previous microinstruction.
Besides the ALU path, several special transfers can be
made to and from RO, R1, R2, and R3 over the T Bus.
When one of these special transfers takes place, with the
exception of loading R1 and R2 from the T Bus, up to four
bytes can be placed on the T Bus or loaded from the T
Bus. These special transfers are as follows:

Transferring data between the


register (BUF) and R03.

Transferring data between the Channel Counter


(CC) and R23.

Transferring data between the Data Transfer


Counter (DTe) and R23.

Transferring data between the four l-byte


registers GO, G 1, G2 and G3 (G03) and R03.

Transferring data
Register and R23.

between

the

Buffer Oata

Transferring data between the Buffer Address


Register (BAR) and G23.

Transferring data between R03 and G03.

G4 Register. The G4 Register is a one-byte register


which can be accessed through the A or BR Bus and can
be loaded from the 0 Bus.
G5 and G6 Registers. The G5 and G6 Registers are onebyte registers which can be accessed through the A Bus
and loaded from the 0 Bus.

When Operational In inbound channel tag is up, bit 5 is


loaded at Register Time by the logical OR of the following
conditions.

PS REGISTER BIT ASSIGNMENTS

4
SR

SR

SR

BR

PAGE

PAGE

PAGE

PAGE

MUX

MUX

MUX

MUX

SEL

SEL

SEL

SEL

SEL

SEL

SEL

SEL

INT REGISTER BIT ASSIGNMENTS

CHNL

OAT';'

BYTE

XFER

2860

CNT

CNT

CNT

MODE

"0

ZERO

ZERO

CHNL
lOX

INT

CHK
RESP

Bit 0, Channel Byte Count Zero. This latch is set when


the SCU Channel Interface has finished its data transfer.
The Channel Byte Counter is decremented after each
byte is transferred, either to or from the channel.
A byte at count one is transferred. When this transfer
has been made, INT Register bit 0 is set and no further
transfers are initiated until the latch is reset. The latch is
reset each time the Channel Byte Counter Register is
loaded.

Subroutine

G03 Register. The G03 Register is composed of the four


one-byte registers GO, Gl, G2 and G3. Each of these
registers can be accessed individually through the A or B
Bus and loaded individually from the D Bus. Additionally,
data may be transferred to or from these registers
simultaneously, utilizing the T Bus. These special
transfers include the following:

3672.21 -0001-10/75

Special Purpose Registers


INT Register. The INT Register is one byte wide and can
be accessed through the A or BR Bus.

latch is gated to Bit 5 at Register Time. The selection


latch is set during the initial-selection channel-sequence,
when the control unit hardware detects that it is being
selected. The selection latch state is not gated to bit 5
when Operational In is set.

Bit 1. Data Transfer Count Zero. This latch is set when


the Data Transfer Counter is decremented from one to
zero. The latch is reset each time the Oata Transfer
Counter is loaded.
Bit 2, 2860 Mode. This bit is set when OPERATIONAL
IN is raised if the channel PCB selected (i.e., CA. CB, CC,
or CO) has the 2860 Attachment Feature installed.

Bit 3, CNT =0. ThiS bit is identical to the B(CNT = 0)


branch condition. It
IS set when the counter incremented from FF to 00 It is reset when the counter
IS loaded (at Register
Time)
Bit 4, Index. This bit is set at register time of an
instruction if Select Alert 2 is active on the controller
interface. This bit is reset by the control statement
C(RESET. INOX) at Register Time.
Bit 5, Channel Interrupt. When the Operational In
inbound channel tag is down, the state of the selection

Interface Disconnect. This occurs when the


channel tag Address Out is set and the channel
tag Select Out is reset.
Truncation. This occurs when the channel
responds to a Service In or Data In tag with the
Command Out tag.

Bit 6, Response. This bit is set at register time of an


instruction if Tag Valid, Normal End, or Check End are
active on the controller interface. The bit is reset at
register time of an instruction if Tag Valid, Normal End,
and Check End are no longer active.
Bit 7. Check 2. This bit is set when a CK 2 condition
occurs. Bit 7 is loaded at each register time from the CK2
latches. CK 2 latch is reset via the C(RESET. CKS) control
statement at Register Time.
IMK Register. This IMK Register is one byte wide and
can be loaded from the 0 Bus. The B(INT) branch
condition is f~rmed by the logical ANO of the INT bits and
corresponding bits set in the IMK Register.
CNT Register. The CNT Register is a one-byte register
which is incremented at Register Time. and which can be
accessed through the A Bus and loaded from the 0 Bus
or T Bus. On any cycle which loads the CNT Register,
incrementing the CNT Register is inhibited and the
Counter Carry latch is reset.
The B(CNT = 0) branch condition is true if the Counter
Carry latch is set. The Counter Carry latch is set at
Register Time, if bits 0-7 of the CNT register are all 1s.
The selected interrupt conditon of CNT = 0 operates
identically to the B(CNT = 0) branch condition.
Page Select Register. The Page Select (PS) Register is
one byte wide, and can be accessed through the A Bus
and loaded from the D or T Bus.

Bits 0-3. Page Select. Bits 0-3 are used to address the
Register File. The Register File word operated upon
during a microinstruction's execution is determined by
the value of the PS Register at the end of the previous
microinstruction's execution. The PS Register is loaded
at Register Time.
Bits 4-7. Branch Multiplex Select. Bits 4-7 are used to
'select the register to be multiplexed onto the BR Bus. The
data on the BR Bus is loaded into a buffer at Register
Time of a microinstruction's execution.
CCK Register. The CCK Register is a multiplex of two
independent registers - the CHC Register and the MCK
Register. MCM Register bit 4 controls which of these two
registers is multiplexed onto the A Bus as CCK. When
MCM bit 4 is a set, the CHC Register is multiplexed onto
the A Bus and the MCK Register is degated.
MCK Register. The MCK Register is the collection of 47
latches which are set by specific error conditions
(paragraph 6.6.1). The contents of these latches can be
accessed eight bits at a time through the A Bus. The
eight bits available at any point in time are controlled by
the MCM Register. The error conditions associated with
each latch are shown in Table 6-3. This table presents
error conditions, along with the necessary states of the
MCM Register, to access the conditions through the A
Bus.

MCM Register. The MCM Register is one byte wide and


can be loaded from the D Bus. The function of MCM is to
control the error conditions available to the A Bus in the
MCK Register. This register also controls the multiplexing of ECR pages to the A Bus, as well as the selection of
A Bus entries from the CHC, SP, and MCK/FDF
Registers. Specific assignments for bits 4-7 of MCM are
also shown in Table 6-3.

6-15

MCM REGISTER BIT ASSIGNMENTS


1

EN

NOT

ECR

ECR

CCKI

MCK

MCK

MCK

CK

USED

MUX

MUX

SFD

MUX

MUX

MUX

BIT

BIT

MUX

BIT

BIT

BIT

BIT

Logical OR (V). Logical OR is performed upon operands


A and B when this function is executed. The carry latch
is reset when this function is executed.

TABLE 6-3. CHECK 1 MULTIPLEX ERROR CONDITIONS

Logical And (.). Logical AND is performed upon


operands A and B when this function is executed. The
carry latch is reset when this function is executed.

MCM REGISTER
BIT POSITION

MCKO

MCK 1

MCK2

MCK 3

MCK 4

MCK 5

MCK 6

MCK 7

CHCPARITY
ERROR

MULTIPLE
WCS

A BUS

B BUS

T BUS

BRANCH

IMPL

ALU

BR

PS

TRANS
LATION

SUB
ROUTINE

WCS ECC

MULTIPLE
BUFF

BUFF
EGG

TBUFFER
ERROR

6.3.5 Buffer Storage


Subroutine (SR) Register. The SR Register is designed
to be loaded with the present microinstruction address
plus one when microinstructions containing the B(LO.
SR. *+1) branch condition or containing the control
statement C(LO. SR. * +1) are executed. Also the SR
Register can be loaded directly into the microinstruction
address register by specifying the branch condition
B(AOR = SRI. In order to provide nesting of subroutines
the SR Register can be transferred to or from R2 and R3
of the selected scratch pad page via the T Bus.

6.3.5.1 GENERAL

P1

P2

P3

P4

P5

P6

P7

P8

SEL

SINGLE

SINGLE

FA 12

FA 11

FA 10

FA 9

FA 8

RESET

BUFF

WGS

FA7

FA 6

FA 5

FA4

FA 3

FA 2

FA 1

FAO

GA

GB

GG

GO

FD
READ

FD
SEEK

FD NOT
READY

WRITE
BUS

The buffer is used to store the following types of


Information:

NOTES
A description of the CC. CI. CHF. CT.
CPo TR. BO and BI Registers will be
found in paragraph 6.2.
A description of the OTC CO. 01. ~O.
and RWC Registers will be found in
paragraph 6.4.

6.3.4.3 ARITHMETIC LOGIC UNIT


Operations
The arithmetic logic unit (ALU) is capable of performing
eight arithmetic or logical functions on two eight-bit data
bytes on the A and B Buses. Figure 6-14 shows a block
diagram of the ALU. The result of the ALU operation is
placed onto the 0 Bus. and for arithmetic operation. the
carry-out is saved in the carry latch. The state saved in
the carry latch can be tested with a branch allowing the
sequence of microinstructions to be altered by the result
of an arithmetic operation. The state of the carry latch is
determined by the carry-out from bit 0 of the ALU. The 0
=0 latch which is utilized by the branching circuitry is set
only when all eight bits of the 0 Bus are zero. The 0 =FF
latch. which is also utilized by the branching circuitry. is
set only when all eight bits of the 0 Bus are all 1 s.

The buffer is a 512-word semiconductor storage unit


completely separate from the WCS with a word size of 39
bits. Thirty-two bits are used for data with the remaining
seven bits being used for ECC checking and correction.
The buffer continuously cycles in a Read operation.
which is interrupted only when a write command is
being executed. Minimum buffer-cycle time for a Read or
Write operation is 480 nanoseconds.

The ALU is actually two ALUs operating in parallel. The


outputs of the two ALUs are compared and. if unequal. a
machine check error condition is set.

execution. This function provides Rotate Left One Bit


Position once the carry latch is loaded or Multiply-by-2 if
the carry latch is zero.

Functions

Compare (K). Arithmetic sum with carry-in of 1 is


performed upon operand A and the 15 complement of
operand B. This function is a true subraction of operand
B from operand A.

Add (+). Arithmetic sum with carry-in of zero is performed on operands A and B. The carry latch is loaded
when this function is executed. and is available for
branching on the next machine cycle.
Add With A Carry (+1). Arithmetic sum with carry-in of
one is performed on operands A and B. The carry latch is
loaded when this function is executed and is available for
branching on the next machine cycle.
Add With a Carry Determined by the Carry Latch (+C).
Arithmetic sum with carry-in equal to the value of the
carry latch is performed on operands A and B. The carry
latch is loaded when this function is executed and is
available for branching on the next machine cycle.
Shift A Left (L). A shift left operation is performed on
operand A with the carry-in latch being set with the
carry-out from the ALU from the previous instruction's

Operation sequence and status for each drive.

Usage and error log counters.

Home Address. Count. and Key records from


Read or Search operations.

Corrected Home Address, Count. and Key


records erroneously read during a Read or
Search operation.

Seek. File Mask, and Set Sector arguments for


disconnected chains when the SCU is operating
in a Multiple Request mode.

Running status for inline diagnostic routines.

Carry latch is always loaded with the carry-out from bit 0


of the ALU when this function is executed. The result of
the comparison (subtraction) is stored in two latches.
which determine if A ;!. B and/or A < B. When this
function is performed. if 0 Bus does not equal all zeros.
the A ;!. B latch is set and the A < B latch is set with the
complement of the carry-out from bit 0 of the ALU. Once
the A ;!. B latch is set. the state of both comparison
latches remain the same until reset vilith the RESET. K
control command C(RESET. K). The comparison latches
can be tested with the branch condition of -COMPARE.
EQ and COMPARE. HI. The -COMPARE. EQ condition is
the compare-unequal latch The COMPARE. HI is the
output of the A < B latch.

T Bus Register

Exclusive-OR (*), Logical Exclusive-OR IS performed


upon operands A and B when this function IS executed

This register is four bytes wide and. during operations


involving the buffer. holds the data which is read on a
Read cycle or written on a Write cycle.

6.3.5.2 DESCRIPTION
A block diagram of the buffer storage is shown in Figure
6-15.

6-16

'-H~~~fl
G5

~I
G6

"

I
I

l____
_
I
G0501:
~

SW BUS

SW 0-7

A-BUS
PRE MPXR

VO
CS

o BUS
RA, VC

'/

DISPLAY MPXR
CS 0-7

VO

AA 0-7

RA

BI~

--

DISPLAY BUS
VD

--

BRANCH CONDITIONS
RA

II'

G0501

~
CHANNEL PRE MPXR
CI

CH A BUS 0-7
FILE 0-7

FILE PRE MPXR


GM
MCK - FDF 0-7
A-BUS MPXR

A1 INT
R03 CARD PRE MPXR
GC
A BUS PRE MPXR 1
GO
A BUS PRE MPXR 2
GO

A BUS

MCK FOF PRE MPXR


EX
~

--

A10-7
A20-7

r'\.

GD500

ALU DECODES
RT

LOB BITS
RE

CONTROL BITS

A
B ALU

B-BUS MPXR

...

A *B

CARRY

FF

G0500

l.....
-~

1----1

o BUS
f\

ALU

ERROR
GE

COMPARATOR
ALU ERROR

G0500

B BUS PRE MPXR 1


GB
B BUS PRE MPXR 2
GA

\.
B BUS
0-7

GEN

RA BITS

RB BITS

BRANCH
CONDITIO~J

A <B

--

PARITY
CHECKER

--

PARITY
CHECKER
G0501

GE

B BUS PARITY ERROR


GE

G0502

G0502

..

A BUS PARIT'f ERROR


GE

Figure 6-14 Buses and ALU Block Diagram

6-17

Buffer Address Register

T BUS' BUF 031

BUFF
PAR
GEN

BUFF
MLX
BUF
BIT
32-38

MB500

,~

T Bur
ERROR

This register is nine bits wide and holds the buffer


address. Loading of this register or incrementing of this
register will initiate a Read cycle on the buffer.

C(G23 = BAR) Command


Error Correction

MD500

NOTE

lit

The buffer storage ECC circuits are


entirely separate from other WCS
ECC circuits in the SCU.

,It

l'
ECC P5-P7
PAR GEN

ECC Pl-P4
PAR GEN

All single bit data errors are detected and corrected by


the ECC circuits. Multiple bit errors are detected but not
corrected. Whenever a correctable data error is detected,
the SCU microprogram is notified by the setting of a
Check-2 error condition (MCK Register bit 1 with multiplex select of 3). Uncorrectable errors will cause a
Check-1 error condition (MCK Register bit 5 with multiplex select of 1).

MD501

BP 1-4
Pl-4

, ,It j

P5-7

BP 5-7

,r

"
ECC
ERROR

BuF~ER

STORAGE

COMPARE
MD501

MB50l
,~

BUF ECC
OUT 0-31

When data is written into the buffer, the ECC is


generated and written into bit positions 32 through 38 of
the storage location. When the data is read from the
buffer, the data is again passed through an ECC syndrome generating circuit. If a correctable error is
detected, three of the seven lines, P1 through P7 will be
high and cause the error to be corrected. -

"
8UFF
MPXR

GL

BUFFER
ADDRESS
REGISTER

C(BAR = G23) Command


Two bytes from general purpose registers G2 and G3 are
transferred to the BAR and a Read cycle is initiated on
the buffer. During the Read cycle, the data from the
buffer storage location . specified by the BAR is
transferred to the buffer ECC. If this command is given
and the buffer had not completed a previous cycle, the
previous cycle is aborted and the execution of this
command initiated. This is a Register Transfer operation
and a Storage-CYcie-Read initiation command.
C(BAR + 1) Command
The BAR is incremented and a Read cycle is initiated on
the BUF. This command has the characteristics of the
BAR = G23 command.
C(R03 = BUF. LD) Command
This is a command which first executes the R03 = BUF
sequence followed by a BAR = G3 sequence. The BAR =
G3 sequence only loads the least signific;ant eight bits of
the BAR.
C(R03 :: BUF. INC) Command

C(R03= BUF) Command

INCREMENT
BAR

Transfers the contents of the BAR to general purpose


registers G2 and G3. Registers GO and G1 remain
unchanged.

6.3.5.3 BUFFER CONTROL COMMANDS


MD502

G3

C(BAR = G3) Command


Transfers the contents of the G3 Register to the BAR.
The most significant bit of the BAR remains unchanged.

BUF MPX
0-31

,r
T-BUS
REGISTER
GA/GB

MB501.

Figure 6-15. Buffer Storage Block Diagram

Four bytes are transferred from the buffer (BUF) to


registers Ra3. The output of the buffer is latched in the T
Bus Register which in turn is gated into Ra3.

This command causes the register-to-register transfer


R03 BUF, a BAR increment (BAR BAR +1), and a Read
cycle initiation. Registers R03 are loaded with the
contents of BUF before BUF is loaded with new data.

C(BUF = R03) Command

C(BUF = R03. LD) Command

Four bytes of data in Ra3 are written into the Buffer


Storage array. The T Bus Register holds the data while
the Write cycle is performed. The location in the buffer is
defined by the contents of the Buffer Address Register
(BAR). The BAR is not modified. If this command is given
before a previous Read cycle is completed, the cycle in
progress is aborted and a new Write cycle is initiated.

This command first executes the BUF


followed by the BAR = G3 sequence.

= R03

sequence

C(BUF = R03.INC) Command


This command first executes the BUF = R03 sequence
followed by the BAR = BAR +1 sequence.

6-18

6.3.5.4 ERROR CORRECTION CODE

TABLE 6-4 BUFFER STORAGE ECC MATRIX

The buffer storage ECC is generated using the matrix


shown in Table 6-4. Bits 0 through 31 of the data word
are examined in each horizontal row of the matrix. Each
1 bit of the data word is examined three times in the
vertical rows of the matrix. If an even number of 1s is
present, one of the P1 through P7 bits is turned on to
make the count for that row odd. This scheme of bit
examination along with comparison of the P1 through P7
bits written into the buffer allows a single data bit error
(bit dropped or picked up) to be detected and corrected.

D
E

Example of ECC Generation

Assume the data word to be written into the buffer as


shown inTable 6-5. Each data bit which is a 1 is applied
to the ECC matrix. For instance, bit 0 which is a 1 in the
example data word is noted in horizontal rows A. B, and
C while. bit 6 is noted in rows A, C, and F, and so forth.

1
1

1
1

1
1

1
1

1
1

P1 P2 P3 P4 P5 P6 P7
1

1
1

1
1

1
1

1
1

12 13 14 1~ 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
1

1
1

10 11

1
1

TABLE 6-5. EXAMPLE DATA WORD APPLIED TO BUFFER STORAGE ECC MATRIX

In each horizontal row 1 bits are counted and if the count


is even the ECC bit for that row is set. For instance, bits
0, 1, and 6 in row A give an even count so that ECC bit
P1 is not set. In row C bits 0, 6, 9, and 19 are an even
count; thus ECC bit P3 is set and so forth.

'5

Example of ECC Correction

Using the matrix in the previous example, assume bit 6 is


misSing (dropped). As data is read from the buffer it is
again examined and compared to the ECC code from the
buffer. Any single bit error will cause three of the lines,
P1 through P7, to be true. In the case of this example,
P1, P3, and P6 will be true, which indicates an error
associated with rows A. C, and F of the matrix. Examination of the ECC matrix shows that the only bit appearing
in rows A, C, and F is bit 6. Therefore bit 6 was in error.
At the proper time, the ANDed outputs of P1, P3, and P6
will be gated into the T Bus Register and the error will be
corrected.

DATA WORD

6 17

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 2930 31

6.3.5,5 BUFFER STORAGE ALLOCATION

The tables shown in Figures 6-16 and 6-16A depict the


Buffer Storage allocation for a typical subsystem.

3672.21-0001-10175

32 33 34 35 36 37 38
P1 P2 P3 P4 P5 P6 P7

1
1

1
1

1
1

1
1

Figure 6-16 shows the Buffer Storage allocation itself


while Figure 6-16A defines the meaning of bytes and
bits within selected words listed in Figure 6-16. Refer to
the Glossary for definitions of abbreviations.

Parity
Parity for the data read from the buffer is generated by
the ECC circuits and gated into the T Bus Register. Odd
parity is used with a parity bit for each byte of the data
word.

E
F

6.3.6 Flexible Disc Interface


6.3.6.1 GENERAL
The Flexible Disc Interface controls the data flow
between the Flexible Disc and the Microprocessor. The
Flexible Disc stores the microprogram and the diagnostic
programs. The Flexible Disc Interface automatically
generates Initial Program Load control signals to the

Microprocessor and to the disc file whenever power is


applied to the SCU.

applied to the FDF MUX, which channels the status via


the A Bus to the Microprocessor.

A block diagram of the Flexible Disc Interface is shown in


Figure 6,17. Two special-purpose eight-bit registers,
Flexible Disc Data (FDD) and Flexible Disc Control (FOG),
are used in the Flexible Disc Interface. Each register is
loaded from the 0 Bus. The FDC Register defines the
data contained in the FDD Register.

Error Correction Code (ECG) circuits in the Flexible Disc


Interface are used to check the ECC from the Flexible
Disc and to generate an ECC for the WCS. The ECC
circuits in the Flexible Disc Interface are used for all
Read and Write operations associated with WCS. Details
of ECC generation and checking may be found in
paragraph 6.3.3.4.

The SERIAL DATA from the Flexible Disc goes to the FD


Shift Register where it is assembled into words, is
applied to the Write Data Buffer, and then strobed out to
the WCS and to the FDF MUX.
The Flexible Disc Interface also contains the logic to
determine the status of the Flexible Disc. The status is

6.3.6.2 FUNCTIONAL DESCRIPTION


The function description of the Flexible Disc Interface
includes a brief description of each functional block,
followed by an IPL (sequence of events) flow chart.

6-19

HEXADECIMAL
ADDRESS

000
I

007

cx:s
I
I

BUFFER STORAGE ALLOCATION

String 0 Drive 0 Data Word 0'

HEXADECIMAL
ADDRESS

BUFFER STORAGE ALl-OCATION

HEXA
DECIMAL
AODRESS

Slri,"g 0 Drif 0 Dala '1ord 3"

07F

String 3 Drive 7 Data Word 3"

10F

String 1 Drive 0 Data Word 0'

080

Slrilng 0 Dri~e 0 Dala "'lord 4"

110

I
I

I
I

I
I

I
I

I
I

I
I

137

Restart Displacement

Log Control Block""

138
139

Retry Parameters

String 0 Channel A Interrupt Word"

128
129

Channel Overrun Counters A and B


Channel Overrun Counters C and 0

Count Buffer

12A
12B

Home Address Buffer

13A
13B
13C

12C

Byte Displacement Accumulated

120

ECC Panern

String 3 Drive 7 Data Word 4"

113

String 0 Channel D Interrupt Word"

010

String 2 Drive 0 Data Word 0'

OAO

String 0 Drive 0 Data Word 5"

114

String 1 Channel A Interrupt Word"


I
I
I

117

String 1 Channel 0 Interrupt Word'

lIB

String 2 Channel A Interrupt Word"

130
I

OBF

Strit"9 0 Dri~e 0 Data "'lord 6"

String 0 Drive 0 Data Word 1


I

I
I
I

I
I
"

llB

String 2 Channel D Interrupt Word'

OEO

String 0 Drive 0 Data Word 7"

l1C

String 3 Channel A Interrupt Word'

OFF

String 3 Drive 7 Data Word 7"

I
I

I
I

I
I

.:

12E

String 3 Drive 7 Data Word 6"

ODF

String 0 Drive 0 Data Word 2"

oeo

040

05F

String 2 Drive 7 Data Word 0"

String 3 Drive 7 Data Word 1 "

String 3 Drive 0 Data Word 0'

03F
I

String 3 Drive 7 Dala Word 5"

Strong 3 Drove 7 Data Word 0"

BUFFER STORAGE ALLOCATION

Counter Unit Check Owed Word'

09F

O~O

HEXA
DECIMAL
ADDR~SS

127

Strong 1 Drive 7 Data Word 0'

:
017

BUFFER STORAGE ALLOCATION

126

OOF
I

HEXA
DECIMAL
ADDRESS

Not Used

100

060

I
I
SIring 0 Drive 7 Data Word 0'

I
I

BUFFER STORAGE ALLOCATION

11 F

I
I

I
I

I
1

17F

Data Buffer

1~0
I

Diagnostic Status and Control Block

18F
Log Sense Bytes 0-23

190

Basic Wait Loop Timer

'~'

Diagnostic Scratch

136

String 3 Channel 0 Interrupt Word"

120

Key Buller

131

I
I

Not Used

130

17E

Sense Bytes 023

IFF

125

Strong 3 Drive 7 Data Word 2"

"See F,gure 6-16A for definitions 01 bytes within word_

FIGURE 6-16.

3672.21-0001-10175

BUFFER STORAGE ALLOCATION AND CONTENTS

6-20

HEX
ADDRESS
RANGE

BYTE 0
0

11

LOGICAL DRIVE ADDRESS

~lrong

~t"ng

~evice

~ev,ce

~ev,ce

I J I I~PS I ~PS I ~hnl I\0

CHANNEL DEVICE STATUS


J2003F

Reserved

Cha,n
In9

Un,t
Check

SIP
or
SSIP

CU 0

I I J
2. '47.
483.
648

I'

.073.
74' .
824

I V

12

256

Index
Error

32 768

I I
cu

1'6.384 I

536
.
870.
912

1268.
435.
456

134

217.
728

167.108133.554'116.777.
864
432
216

8.192

8.388.
608

4.096

2.048

1.024

1 1 .024

512

256

128.

64

SEEKS

'NOT USEDI

{NOT USEDI

POLL ENABLE. STRING 0


12

POLL ENABLE. STRING 1


12

POLL ENABLE. STRING 2


0

\1

4.096

2.048

512

256

[IJ

I'

1 3

\1

32

16

Online

1 4

15

17

\ 7

\ 2

2.097.
152

1 048

576

524.
288

64

l'

l'

\ 4

\5

II

0
3

17

I'

32

16
1

1 1 I
5

'

32

1 2

-'

I
\

BYTE 3

I I
4

D
5

Anentlon

Busy

_14

12

262
1 144.

131
072

PACK CHANGE DE OWED

I I
C

I I I
B

11

12

65.
536

.32.768

16.38418.192

1 16

32.768

'6.384

16

4.096

2.048

8.192

1 024
.

ill

512

.1

I
I
I

SM

CUE

Seek

Mask

Busy

I I 1

128

64

Seek

Mask
2

17

11J
Diag

AI!

Write

Auto

Mask

Pad

PCI
Fetch
Mask

CE

16

DE

UE

uc
1

1 1

Double!
Density

32
1

14

12

11

RETRY DATA CHECKS


4.096

2.048

1.034

512

256

128

64
1

128
I

164

(NOT USED)

32

'6

32

16

12

I'

11

15

17

SEEK ERRORS

[II

DEVICE END READY. STRING 0

I,

DEVICE END READY. STRING,

I'

12

3
1

DEVICE END READY. STRING 2

11

I' I

0
3

11

UNIT CHECK OWED. STRING 3

I'

DEVICE END READY. STRING 3

UNIT CHECK OWED. STRING 2


\

I 1

(NOT USED)

15

Cham

{NOT USED,

In.

(NOT USED,

[II

Mask

INOT USED)

SFM

Wrote

BYTES READ
256

INOT USED)
5

Stacked

11

(NOT USED)

15

I I

DEVICE TYPE

12

{NOT USEDI

ill

[II

Mask

".

12

CHANNEL STATUS

HEAD ADDRESS

1512 DO 1256 DD
256 SO

'

11

Write
A

BIT SIGNIFICANT DEVICE ADDRESS

Sector
Ready

FILE MASK

(NOT USED)

UNIT CHECK OweD. STRING 2

RETRY DATA CHECKS

DISCONNECT COMMAND CHAINING. STRING 3

11

BYTES READ

DISCONNECT COMMAND CHAINING. STRING 2

m
I I
2

128

plete

FIGURE 6-16A.

3672.21-0001-10175

DISCONNECT COMMAND CHAINING. STRING 1

UNIT CHECK OWED. STRING 0


,26

DISCONNECT COMMAND CHAINING. STRING 0

POLL ENABLE. STRING 3


llC-llF

CORRECTABLE DATA CHECKS

INOT USED,

118-11B

HIGH CYLINDER

4.194.
304

OEOOFF

116.3841 8.192

11

11

Seek
Com-

I I

32.768

DE OWED

Seek
Incom
plete

tNOT USED,

114-117

64

oeOODF

I'

Offset
Actrve

0A0-06F

BYTES READ

SEEKS

110'13

I I I I I I I

,2B

CORRECTABLE DATA CHECKS


08009F

12

DRiVE STATUS 1

BYTES READ
060-07F

11

LOW CYLINDER

HIGH CYLINDER
04005F

LAST PHYSICAL ADDRESS

I I I I I I I

ooo-OIF

BYTE 2

BYTE 1

[II

0
3

BUFFER STORAGE BYTE DEFINITIONS (1 of 2)

6-20A

BYTE,

BYTE 0

HEX
ADDRESS
RANGE

' I

Im I I I I I I ,

LOG CONTROL BLOCK

127

UC.
Log 510

Oone
Log

S",ng
4

Smng
2

5'''"9
1

Dnve
4

Dr,ve
2

Dflve

RETRY SECTOR VALUE

138

64

.28
1

32

128

64

32

I , I

BYTE 2

I I I I I I I,

LOG CONTROL BLOCK

Ms~

Ms~

'6

12

I'

16

128

64

I'

1 32

BYTE 3

1'6

1 8

I
,28

1 J

I
64

I ' I

11 1 1 1 I

LOG CONTROL BLOCK

I I I I I

I'
1

LOG CONTROL BLOCK

DATA DVERRUN COUNTER

[I]

1, I

I
,

{NOT USEDI

(NOT USED)

Mssg

Fo'ma' 4
Fo,ma' 2
Fo,ma' 1
Fo,mat8 4
Mssg
8
2

CURRENT OFFSET

COMMAND OVERRUN COUNTER

'39

32

16

J
I

Log

Sense

Log

Log

Count

Count

DATA CHECK RETRY COUNT

128

64

32

16

16

I'

SEEK RETRY COUNTER

2
1

128.1

64

32

/,

When bit is set to a 1. numbered device is available to this Channel.

When bit is set to a 1. numbered device has Disconnect Command Chaining in progress for this channel

III ~:e~~~\ I~:~~~ ~ 1~ ~~~ec~:~~~ ~~:~r~p:h:~~~~.~;:1 ~:~~ a previously presented

~hen

IT]

Bit set to a 1 Whe'n

co~nt equals 4. reset to a 0

FIGURE 6-16A.

3672.21-0001-10175

Device

Busy status (generated by

bit IS set to a 1 Unit Check is owed for this device


when logging IS Initiated

BUFFER STORAGE BYTE DEFINITIONS (2 of 2)

6-208

I
I

FLEXIBLE
DISC

I
I
I
I
I
I

READ
LOGIC

I
I

..
.....
..

LOAD MEMORY

FLEXIBLE DISC INTERFACE

WRITE ENABLE

SCAN

IPL
ADDRESS
GATE

WRITE ENABLE

FC510

.....

WCS DATA (RBI

..

SEP DATA

D-BUS (GDI

FDD MUX
AND
REGISTER

FDD BITS

--..

....

FC520

....

SEP CLOCK

..
F

FDC MUX
AND REGISTER

I
I

FD
SHIFT
REGISTER

SECTOR

~ ~

I
I

..

-...

I
I
I

....

WRITE
DATA
BUFFER

....

~
SEP CLK

ADDRESS
REGISTER

...

EOB, TRK 00

.....
..
r

READ
LOGIC

...

~ ~

DECODER

..

FC520

IPL
LOGIC

IPL STEP OUT


FC560

(GEl'"
FD560
IPL ADDRESS BITS--.
(RA)'"

FC500

can be found in Section 5 of this manual.

RD
STATUS

."

...

FD NOT READY
POWE RON-RESET

FDF ..

FDF
MUX

Manual. Interconnections to the Flexible


Disc unit are also shown. Details of the
Flexible Disc unit theory of operation

READ GATE

.....

...

NOTE
The block diagram on this page should
be used in conjunction with the theory
of
operation
on
Flexible Disc
the
Interface. The cali-outs noted in the
lower right corner of each block refer to
logic page numbers in the Logic Diagram

SET POINTER

TRK 0, SECTOR

TOWCS ..

FC590

FC595 I""
~

r-

:..

..

FD530
FD535

..

.......

ECC
GENERATOR
AND
COMPARATOR
FD580
FD585
ECC
BITS

ENABLE 6, 7

...
....

TO WCS ..

.. ENABLE BR + POINTER

FC520

FD540

POINTER
AND
BIT RING
COUNTERS

5 -7

CONTROL
LOGIC

ENABLE 0 - 5

INCR. ADDR. REG

~ ~ ~1~TROLS

....

SHIFT DATA

FLEXIBLE
DISC
MECHAN-ISM

.....

CLEAR DATA REGISTER

I
T,READ
DATA

TOWCS

STEP IN (OUT) CMD

---.....

SEEK STATUS

SEEK
LOGIC

FD
STATUS
LOGIC

...

READ AND SEEK ERRORS ..


FC580

(GE) ...

FC570

Figure 6-17. Flexible Disc Interface Block Diagram

6-21

FDD Register

Data bits 5, 6, 7 of FDD will be transferred into


the pointer (FDP). The pointer values correspond
to the following:

The FDD Register is loaded from the D Bus. The output of


the register is routed to the FD Shift Register, the FD
Pointer, or the Address Register. The destination of the
FDD Register output is controlled by the FDC Register.

FDP
FDP
FDP
FDP
FDP
FDP
FDP
FDP

FOC. Register
The FDC Register is loaded from the D Bus by raising FD

= FDC. The output of the FDC controls the FDF MUX, the
Seek LogIc, the Read Logic, the Address Register and the
FO Pointer which in turn controls the data strobed
through the Write Data Buffer.

FOF

FOF

FOF

MUX

MUX

MUX
2

POWER ECC
BIT
ON TO
FO
SELEC
TION

FOC

FOC

FOC

MUX

MUX

MUX
2

FDC Reg. bits 0-2; FDF MUX. These three bits enable
reports through the FDF multiplexer. Loadedfrom the D
Bus by raising RD = FDC.
FOC Reg. bit 3; Power On to FD. When set to 0, this bit
enables the FD to be powered on. When set to 1, this bit
enables the FD to be powered off. Loaded from the D Bus
by raising RD = FDC.
FDC Reg. bit 4; ECC Bit Selection. When set to 0,
enables ECC bits generated by the FD interface logic to
be loaded into WCS. This bit is set to
during normal
operation. It is set to 1 only during error diagnostics.

FDCMUX =

FDCMUX = 1 -

Set Pointer

32-39 of Write Data Buffer


40, 41 of Write Data Buffer
0-7 of Write Data Buffer
8-15 of Write Data Buffer
16-23 of Write Data Buffer
24-31 of Write Data Buffer
0-7 of Address Register
8-11 of Address Register

FDCMUX = 2 -

Set Data

FDCMUX = 3 -

Start Micro Read

FDCMUX

=4

Step In

The FD will move the access mechanism one


track away from track 0. This condition is
latched up and is reset with SEEK COMPLETE.

FDCMUX = 5 -

Step Out

The FD will move the. access mechanism one


track toward track 0. This condition is latched up
and is reset with SEEK COMPLETE.

FDCMUX = 6 -

Reset Word Ready

De-activate the word ready indicator. Failure to


execute this instruction within 38 microseconds
after Word Ready indicator is activated will
result in an overrun condition.

The Read logic is the hardware control for loading the


bootstrap section of IPL. This logic also controls the
loading of the rest of IPL and diagnostics by way of
control signals from the Microprogram.
FD Seek Logic
The Seek Logic causes the FD to automatically seek to
track QOO during bootstrap and also to step in or out to
any track called out by the microprogram.

The IPL (Initial Program Load) Logic and read Gate


contains the logic required for initial program loading.
An IMPL error is detected in the IMPL Error Logic and is
an indication of the following:

FDCMUX = 7 -

Release

The FD interface is available to the second


processor of a dual processor machine, This
does not cause the FD to power down.

at a time, a complete data word (42 bits) into the Write


Data Buffer. The output, Address Bits 0-1 1, are routed to
the WCS Address Register and to the FDF MUX.
Address Register
The Address Register contains the address where data
from the FD, will be loaded into memory. During IPL, th~
Address Register is incremented for each data word,
starting at address 000. For other operations the address
comes from the FDD Register.
Write Data Buffer

IPL Logic

The Write Data Buffer temporarily stores, and makes


available to WCS, the data to be written into memory. It
also stores data out of memory during a read control
store. The output is also available to the A Bus through
the FDF MUX.

Read Overrun as described in FDF bit 5.

FD Status Logic

Seek Time Out one second.

The FD Status and IMPL Error Logic accepts signals from


throughout the Flexible Disc Interface logic circuits, and
generates ready or error signals which are sent out to
the Microprocessor via the FDF MUX.

Seek not completed within

Read Time Out as described in FDF bit 5.

Sixteen occurrences of either or combinations


of the following:
a. IPL Seek Error - no track =
two sync bytes during IPL.

decode after

b. Data Bus Parity - parity error detected


while reading from the FD.
FD Shift Register
The FD Shift Register takes serial data from the flexible
disc and assembles it into bytes to be loaded into the
Write Data Buffer a byte at a time. The Shift Register can
also accept a parallel load of one byte of data from the
FDD register.
Pointer And Bit Ring

No operation.

bits
bits
bits
bits
bits
bits
bits
bits

The Read logic will be activated and data will be


transferred from the FD to the Write Data Buffer
until EOB is received. This condition is latched
up and is reset with either END-OF-BLOCK or
RELEASE.

FDC Reg. bits 5-7; Decodes. These bits go to a BCD-todecimal decoder from the D Bus which generates eight
decodes. The eight decodes are defined as follows:

0:
1:
2:
3:
4:
= 5:
= 6:
= 7:

The data in FDD will be loaded into the area


corresponding to the pointer value.

FOC REGISTER BITS

=
=
=
=
=

FD Read Logic

The pointer points to a particular section of the Write


Data Buffer (enable 0-5) into which a single byte of data
will be loaded. During bootstrap IPL, the bit ring counts
bits of data off the Flexible Disc and increments the
pointer for each byte of data to sequentially load, a byte

ECC Circuit
The ECC generator accepts data bits during a read
operation and generates anEC code. The code is
compared with the ECC initialiy written into the disc
during a write operation. If the ECC compares, the code
is written into memory either from the ECC generator or
from the Flexible Disc, depending on the state of bit FDC
4. A noncomparison results in a Read Data Check signal
to the FD status logic.

FDF MUX
The FDF MUX makes data available to the A-Bus. This
data can be either Write Control Store data, IPL Address
Bits, or Status Bits denoting the status of the FD
interface. The data is multiplexed through a byte at a
time by FDC bits 0, 1, and 2. A chart depicting the FDF
MUX output bytes is shown in Table 6-6. A description of
the FD status bits (FDC 0, 1, 2 = 0) is provided in the
following text. 3 OO
FDF MUX = 0, FDF Bit 0 (Overrun). Word ready was not
reset before new data was loaded into the Write Data
Buffer.

6-22

TABLE 6-6. FDF MUX OUTPUT BITS


FOF MUX
OUTPUT BITS

FOC 0, 1,2

=0

OVERRUN

SEEK
COMPL

FOC 0, I, 2

=1

A7

A6

FOC O. I, 2

=2

FDF MUX
following:

FO
SEEK
READY

FO
READ
ERROR

FO
NOT
READY

BUSY

WORD
READY

A5

A4

A3

A2

Al

AO

1\12

All

10

A9

A8

36

37

28

029

EOB

40

0 41

only)

FOCO,I.2;3

0 32

0 33

0 34

FOC 0, 1,2

=4

0 24

FOC 0, I, 2

=5

017

FOC 0, I, 2

=6

FOC 0, I, 2

=7

DO

O
2

16

25

0 42

26

18

10

FDF MUX
0, FDF Bit , (Seek Complete), SEEK
COMPLETE is normally up unless a Seek is initiated by
raising STEP IN or STEP OUT. SEEK COMPLETE will
come back up, allowing another Seek to be initiated,
within a minimum of 20 milliseconds and a maximum of
40 milliseconds after a Seek is initiated.

35

19

0 21

011

20

FDF MUX 0, FDF Bit 4 (FD Seek Error), An FD Seek


Error is an indication of the following:

A Track = 0 Decode was not found after two


successive Sync Byte decodes during IMPL.
A Seek was initiated and was not completed
within one second.

AO through A 11: Address of Writable Control


Storage into which write bus data will be
written.
DO through 041: Write bus data.

12

13

30

22

14

The FD track format, depicting Sync Bytes, Record 10


Bytes and Data Bytes is shown in Figure 8-16.

39

6.3.6.3 IMPL FLOW CHART

31

23

The IMPL flow chart of Figure 6-19 depicts the sequence


of events which occur in the Flexible Disc Interface logic.
The flow chart also shows timing (i.e., leading edge of

Sector Pulse), and references the appropriate logic


diagram page number.

6.3,7 Microprocesso.r Timing


Microprocessor timing for the SCU is generated by a 25MHz crystal controlled oscillator. The output of the
oscillator is fed to a six-stage ring counter whose
outputs, A through F, furnish the baSic timing pulses for
the SCU. See Figure 6-20.
The nominal timing shown in Figure 6-23 and detailed
timing shown in Figures 6-22 and. 6-21 show the
relationship of basic machine timing to major timing
signals. Each major timing pulse is 40 nanoseconds wide
resulting in a complete timing cycle of 240.
Except where noted on the timing diagrams. the trailing
edge of the clock pulses are used to set the logic.

15

192 BITS
(ALL O'S)

2 SYNC
BYTES

DATA

RECORD
10 BYTES

(70 WORDS MAX)

o0

FDF MUX 0, FDF Bit 5 (FD Read Error). An FD Read


Error is an indication of one of the following conditions:

A Data Bus Parity error was detected during a


read from the FD.

An overrun condition occurred; i,e., WORD


READY is still up when the next byte of data
from FD is ready to be loaded into the Write Bus
Data Buffer during a read.

HEXADECIMAL

TRACK
NO.

00

FDF MUX 0, FDF Bit 3 (Word Ready), WORD READY


is active when the data on the write bus is stable and
ready for loading into the Writable Control Store.

38

27

FDF MUX 0, FDF Bit 2 (Busy), After executing an FDC


= 00 the busy bit will be on if the FD Interface is
connected to the other processor of a dual processor
system.

through 7. These bits control the

Track Format

(Setr 2

A read from the FD was initiated and a sync


byte (32, 32) was not detected in the read data
within one second.

HEXA
DECIMAL

SECTOR
NO.

00

20

040,041,111111
NOTE

LAST BYTE (BITS 2-7) MUST


ALL BE 1'5 TO OBTAIN EOB
DECODE.

40
60

FDF MUX =, FDF Bit 6 (FD Not Ready). Power on


signal is activated to PDU but FD is not operational. This
line is up normally for 5 seconds.

80

AD

FDF MUX
0, FDF Bit 7 (EOB), The following
conditions will reset EOB (End of Block):

CO
31

A machine reset
Initiating a Microprogram read
Release

49

EO

Figure 6-18. FD Track Format

6-23

r--~

STEPS TO STEP
OUT IINI STATE 2
fFC5701

NOTE
GUARANTEES PROP
E R TIME BETWEEN
CONSECUTIVE STEP
IN (OUTI PULSES

(2 OF 3)

LEADING EDGE OF
SECTOR PULSE

IMPL
SWITCH

POWER ON RESET

L
f'

SETS IP~ LATCH


rFC,,60,
ENABLES PWR 00;
TO FD
IDE GATES ALL
INPUTS TO ADDR
REG EXCEPT IPLI

.
I

LEADING EDGE OF
SECTOR PULSE

REG TIME

B7 OF 2ND SYNC BYTE

STEP TO IPL ADDR


GATE STATE 1
(FC510)
IPL INHIBIT AND
IPL ADDR GATE
ACTIVE TO ppROC.

YES

STEP TO READ
STATE 2
IPL SEEK ERR
(AFTER 16 PASSES
THRU HERE IMPL
ERR WILL BE
ACTIVATED)

LEADING EDGE OF
SECTOR PULSE

TRAILING EDGE OF
SECTOR PULSE

*IPLINHIBIT
*IPL ADDR GATE
*IPL WR ENABLE

*ACTIVE TO PPROC
(+WORD WRITTEN
INTO WCSI
REG TIME
STEP TO IPL
ADDR GATE STATE 2
*IPL INHIBIT
*IPL RST WORD RDY

STEPS TO STEP OUT


liN) STATE 3
RESETS STEP OUT
(INI LATCH IFC5701

~~~~~~GpG~~EE

REG TIME
STEP TO IPL ADDR
GATE STATE 3

IPL RST WORD RDY


RESETS WORD
READY LATCH
(FC5801

STEP TO READ
STATE 7
ENABLE BIT
RING CTR
ENABLE BYTE
POINTER
ENABLE 0 ACTIVE,
TRK :!i:LOADED
INTO DATA REG
(FD530,535)

SET READ GATE


LATCH (FC5601

ENABLE 1 SETS
WORD READY
LATCH (FC5801

REG TIME

STEP TO READ
STATE 3
ENABLE BIT
RING CTR

STEPS TO STEP OUT


(IN) STATE 1 (FC570)
STEP OUT (IN) SENT
TO FD INTERFACE

LEADING EDGE OF
SECTOR PULSE

RESETS IPL STEP


OUT LATCH (FC5601

SET IPL STEP OUT


LATCH IFC5601
SET STEP OUT
liN) LATCH
(FC5701

STEP TO READ
STATE 1 (FC500)
ENABLE BIT
RING CTR
CLEAR BYTE
POINTER
CLEAR ADDR REG

RESETS TO STEP
OUT liN) STATE 0
CONDITIONS SEEK
COMPLETE SIGNAL
(FC570

STEP TO READ
STATE 5
ENABLE BIT
RING CTR
ENABLE BYTE
POINTER
ENABLE 1 ACTIVE,
SCTR
LOADED
INTO DATA REG
(FD530,5351

STEP TO READ
STATE 6

OF .... '-'

(2 OF 3)

Figure 6-19. IPL Flow Chart (1 of 3)

(1 OF 3)

RD = FDC (START
SETS READ
GATE LATCH)
FC520 ,FC560,F 0550
(ALSO SET UP
FDC 0-3 TO
USE FDF BIT 3
FOR MONITORING
WORD READY)

~READ

ENABLE 1 ACTIVE,
FD DATA BYTE S
LOAD INTO
DATA REG

RESET TO IPL
ADDR GATE
STATE 0

RD = FDD (SETS FDD5,


S,7) FC520

87

RD = FDC (SET
DATA ACTIVATES
ENABLE 6 LOADING.
FDD 0-7 INTO
ADDR REG 7 -0)
FC520,FC595,FC590

ENABLE 2 ACTIVE,
FD DATA BYTE 1
LOAD INTO DATA
REG (FD530,535)

NO

B7

ENABLE 3 ACTIVE,
- FD DATA BYTE 2
NOTE
J.lPROGRAM
B7

ENABLE 4 ACTIVE,
FD DATA BYTE 3,
INCR IPL ADDR
REG

READ

RST TO READ
STATE 0 (FC500)
RST IPL LATCH
(FC5S0)
-ENABLE ADDR REG
BEGIN J.lPROG
EXECUTION

(1 OF 3)

RD = FDD (SET FDD


5-7)

SE~UENCE. ALL
~PROGRAMS WHICH

B7 AND SYNC BYTE

REOUIRE DATA.
FROM F 0 BEGIN
HERE (I.E. DIAG
NOSTIC LOAD, COM
P LETION OF IMPL).

RD = FDC (SET POINT


ER LOADS FDD5-7
INTO BY BYTE POINTER,
ENABLE 7)
,

B7

ENABLE 5 ACTIVE,
FD DATA BYTE 4

STEP TO READ
STATE 3
ENABLE BIT
RING CTR

B7

RD= FDD (SET


FDD 0-7
FOR WCS ADDR
BIT 7 THRU 0)

RD = FDD (SET
FDD 4-7 FOR
WCS ADDR
BIT 11 THRU 8)

RD = FDC (SET
POINTER LOADS
FDD 5,S,7 INTO
BYTE POINTER,
ENBS) FC520,FC595

RD = FDC (SET
DATA ACTIVATES
ENABLE 7 LOADING
FDD 4-7 INTO
ADDR REG 11-8)

B7

ENABLE 0 ACTIVE,
FD DATA BYTE 5

STEP TO READ
STATE 1
ENABLE BIT
fliNG CTR
CLEAR BYTE
POINTER FC500

STEP TO READ
STATE 7
ENABLE BIT
RING CTR
ENABLE BYTE
POINTER
ENABLE 0 LOADS
DATA BYTE INTO
DATA REG
(FD530,535)

(3 OF 3)

B7

Figure 6-19_ IPL Flow Chart (2 of 3)

6-25

(2 OF 3)

BYTE POINTER STEPS


THRU 2,3,4,5,0 THEN
1 LOADING APPROPRIATE DATA BYTE
FROM FD INTO
DATA REG.
ENABLE 4 INCREMT
ADDR REG

STEP TO READ
STATE 5
ENABLE BIT
RING CTR
ENABLE BYTE
POINTER
ENABLE 1 LOAD
DATA BYTE INTO
DATA REG
(FD530.535)

25 MHZ
XTAL

~....-

- - - - -__~

OSC

'/,
CLK

'!, OSC CLK

TIME
TOGGLE

ENABLE 1 SETS
WORD READY
LATCH TO jJPROC
THRU FDF BIT 3

A TIME

B TIME
C TIME

RST READ GATE


LATCH
RST TO READ
STATE 0
END OF SEQUENCE

jJPROC SENDS WRITE


ENABLE WHICH (FC510)
LOOPS BACK TO )JPROC
TO WRITE WORD INTO
WCS

o TIME
E TIME

F TIME

RD =FDC (RESET
WORD READY
RESETS WORD READY
LATCH)
FC580

LOGIC ONRE501

COMPLETE IPL
ACTION OF READING ONE RECORD
OF DATA FROM
TRACK 0, SECTOR O.
1ST PASS DATA
BYTE IS TRACK
NUMBER
1ST PASS D.ATA
BYTE IS SECTOR
NUMBER

Figure 6-19. IPLFlow Chart (3 of 3)

Figure 6-20. Microprocessor Timing Logic

6-26

T240iO

I
I

T40

I
I

MACHINE TIME (RE50l)


F TIME (RE50l)

~~------------------~~~----------

A TIME (RE50l)

__________~r___l

B TIME (RE50l)

____________~r__l

C TIME (RE50l)

r---I~

________ ________
~

__

__~--

r---L-

r-

~r---l

__________________

~r___l~

___________________

D TIME (RE501)
MEMORY LATCH TIME (RE501)
MACH RESET (RE501)

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _..J~

DATA BITS LATCHED

J4: FIRST "A" TIME AFTER E. O. B. IS ACTIVE


____________
____________________
~r___a~

ADDRESS REG RESET (RE501)

~~

ADDRESS REG SET (RE50l)

L...-_ _ _ _ _ _ _ _---Jr

WRITE ENABLE (WCS) (RE501)

______________________J~~~~SE~T____________________~r_
SET INCREMENTER (RE501)

____________________________---J~~~
__S~T~E_P_____________________
STEP INCREMENTER (RE501)
~I.~~~

T REGISTER CLOCK (RE501,

o REGISTER

CLOCK (RE501)

REGISTER TIME (RE50l)

______________

-fl--------~
~~

ClK FOR CT REG IF CT DECQDE IS"'----IN C" FIELD OF MICROINSTRUCTION

________________ ________
~

BR CLOCK (RE50l)

____

~r___I~

____________r____t

____________

~r---l~

________________________

______________

~r____1~

___________

CO CLOCK (01500)
DO CLOCK
RESET CHK 1 (GE5OO)

____~r____l

________~r__J

.r---I~

____________

r___l.~

_____

CHECK 2 (GE5OO)
MULTIPLE OR SINGLE
WCS ERROR (GE5OO)
RBC ERROR CLOCK
EARLY ERRORS
LATCHED (GE50l)
LATE ERRORS
LATCHED (GE50ll

__~-----------------J~----------------------------L
~r-------------------------~---------------______________~_2~r--------------------------

Figure 6-21. Nominal Microprocessor Timing

6-27

T240/0

NOTES

1. Shaded areas on each side of the


waveforms indicate allowable timing
variations under worse case conditions.

T40

T80

Tl20

T160

T240/0

T200

2. Reset always follows set on leading


edge by;S; 10 nsec and falls;S; 10
nsec before set falls.

I
I

J
J

MEMORY
LATCH TIME
211

194
REGISTER
TIME

10

21

211

194

I
76

106

178

10

206

D REGISTER
CLOCK

T REGISTER
CLOCK

97

112

137

152

155

190

007

213

INCREMENTER
STEP

---144

110

INCREMENTER
SET
194
BRANCH CLOCK

ADDRESS
REGISTER SET
ADDRESS
REGISTER
RESET

206

234

I
194

006

42

73

I<
" 57

r,
1
'

88

122

110

120

I I

Figure 6-22. Detailed Microprocessor Clock Timing

138

206

21

234

006

T40

CLOCK

WCS DATA REG GOOD

WCS ADDRESS GOOD

Tao

T120

T160

T200

T40

T240/0

Tao

T120

ij

;,>

WCS OUTPUT GOOD

GOOD WCS DATA (ECC


COMPLETED)
BRANCH CONDITION LOB
BITS GOOD

a. SCU Addressing

I
I

I
I

I
J
o

BUS GOOD

ALU INPUTS GOOD

ALU BRANCH CONDITIONS


GOOD

/
1

I
I

<,

2'1

..I

I
I

".

........ \.

l .. i/.i
.;1

. .. . .. .. . >

>.

k ...

......:

.I . . .

I
I

'-

;;:::..

:?
I

ALU DESTINATION REG


DECODES GOOD
"FROM" BUS DECODES GOOD

"TO" BUS DECODES GOOD

b. ALU Path Timing

Figure 6-23. Detailed Microprocessor SCU Addressing and ALU Path Timing

6-29

6.4 CONTROLLER INTERFACE


6.4.1 General

BUS

BUS

The Controller interface (CTl-l) consists of specialpurpose registers. counters and support hardware
organized to perform the sequencing and data transfer
functions between the SCU and the Controller. A block
diagram of these elements is shown in Figure 6-24_ The
CTl-1 is under direct control of the SCU Control Program.
Data transfer operations within the SCU are initiated
when the CPU transfers a command to the SCUrequiring
such a transfer. Data transfers between the Controller
and the SCU are initiated by read and write commands:

CTl SYNC OUT

.....

BUS OUT

..
.

CTl TAG GATE


CTl RESPONSE
CTl RECYCLE
LINE
DRIVERS

CE COMMUNICATION
~

S
T

Read commands are those commands which


cause data to be transferred from the Controller
to the SCU.

CTl TAG OUT O. 3-7. P

0
R
A
G
E

------

CTl SELECT HOLD

a. Read Commands

------

CTl BUS OUT 0-7. P--

b. Write Commands
Write commands are those commands which
cause data to be transferred from the SCU to
the Controller.

TAG OUT
TAG GATE
SEl HOLD

BUS

0
N
T
R

During read and search operations. read data bits are


transferred from the selected Disc Drive through the
Controller to the SCU. The data bits are serially
transferred from the Disc Drive to the Controller. The
Controller transforms the serial data into parallel by
bit/serial by byte data and transferred to the SCU for
subsequent transfer to the System/370 Channel or for
comparison with data from the System/370 Channel.
The Microprocessor has access to the assembled data
bytes when the control program addresses the Data
Buffer Register.
logic for the CTl-1 is located on two PCBs. The DC board
contains the major portion of the CTl-1 circuitry such as
registers. error checking logic. and line drivers and
receivers. The GM board contains the general multiplexIng logic to route data from a selected CTl-1 register to
the AlU via the A bus. This board also contains the Data
Transfer Counter.

BUS IN

0
N
T
R

0
l

CTl BUS IN 0-7. P

---

---

-"

---SYNC IN

o
N

T
R

----

LINE
RECEIVERS

NORMAL END
UNSEl ALERT 1

SELECT ACTIVE

---

TAG VALID

"

---

SYNC IN
SEL ALERT 1
CHECK END
SEL ALERT 2

ALL lOGIC ON DC BOARD

RECYCLE

DTC = 0

6.4.2 Interface Signals


Data and control signals transmitted between the SCU
and Controller are buffered in the SCU by means of the
line drives and receivers shown In Figure 6-25.

Figure 624. Controller Interface Block Diagram

Figure 6-25. Storage Control Interface


Line Drivers and Receivers

6-30

6.4.2.1 SCU TO CONTROLLER SIGNALS


Bus Out
The Bus Out lines are used for two purposes. The lines
carry tag modifier, control, or address information when
deskewed and validated by TAG GATE. The lines transmit
data from the SCU to the Controller when deskewed and
validated by SYNC OUT. The Bus Out lines consist of
eight data lines plus parity.
Tag Out Bus
The Tag Out bus sends the five-bit tag instruction to the
Controller to identify the operation to be performed. TAG
GATE validates the instruction on the Tag Out bus. The
bus consists of six bits plus parity; however. bit 3 of the
bus is not used by the SCU or Controller.
Odd parity on the Tag Out bus is required. Even parity
inhibits the TAG VALID signal. thereby preventing the
Storage Control from communicating with the Controller
and the Drives.

Select Hold

CE Communication

Tag Valid

The SELECT HOLD line rises during any select tag and
remains up to maintain selection of a Controller and/or
Drive. The line stays up until the end signal of the last
operation to be performed on the Controller and/or Drive
is received and acknowledged (see Figure 6-26 for timing
relationship).

The CE Communication (CE COMM) line is used for


diagnostic purposes only.

The TAG VALID line indicates that the Controller or Drive


has validated and accepted the tag instruction sent from
the SCU.

Select Active

Sync Out
The SYNC OUT line is used to validate and deskew the
Bus Out bits during data transfers from SCU to Controller. It is also used during data transfers from the
Controller to SCU to check the data count (see BUS
OUT/SYNC OUT waveforms of Figure 6-27 for timing
relationship).

The TAG GATE control line is used to deskew the Bus


Out and Tag Out lines to the Controller and Drive. TAG
GATE is also used to establish the time at which the Tag
Out and the Bus Out lines are valid. The line remains
valid until acknowledged by the Controller with TAG
VALID (see Figure 6-26 for timing relationship).

The SELECT ACTIVE line becomes active as a result of a


selection sequence. The line remains active to indicate
proper selection as long as SELECT HOLD is active and
selection of the Drive is correctly maintained by the
Controller.
Sync In

Response
This line indicates acknowledgement of a Normal End or
Check End condition for extended (Read/Write)
operations.
Recycle

Tag Gate

6.4.2.2 CONTROLLER TO SCU SIGNALS

The RECYCLE signal is used during a (read or write) data


transfer to keep the four-bit data transfer counter in the
Controller counting when more than 16 bytes of data are
to be transferred. The signal is controlled by the data
transfer counter in the Storage Control. RECYCLE
prevents the Controller from setting End of Data Transfer
until the SCU drops RECYCLE.

During data transfers from Controller to SCU, SYNC IN


validates and times the Bus In data. It rises after Bus In
data is valid. Bus In data remains valid until after SYNC
IN ends. During data transfers from the Storage Control
to the Controller, SYNC IN provides timing for data
transferred (refer to BUS IN/SYNC IN waveforms of
Figure 6-27 for timing relationship).
Normal End
NORMAL END indicates that the normal ending of an
operation occurred with the expected results obtained.
Ending information on Bus In is validated by the rise of
NORMAL END. For immediate tags, NORMAL END is
generated by TAG GATE, TAG VALID, and NOT DATA
TRANSFER. NORMAL END drops when TAG GATE goes
inactive.

Selected Alert 1
This line indicates an unusual condition (Equipment
Check) in the selected Controller or Drive. This line is
Reset by a Check Reset or Controller Reset operation
performed by the SCU.
Selected Alert 2
The SELECTED ALERT 2 line indicates the detection of
Index in the selected Drive. This line will not be active
unless a Drive is presently in operation (i.e., an Operate
Up tag (decode 8B) has been issued after selection).
Unselected Alert 1
One line used for diagnostic purposes only.
Bus In
Data from the Controller or Drive during a Read operation, as well as ECC conditions or error information, are
transmitted to the SCU by means of the Bus In lines. This
bus consists of nine lines: eight bits plus parity.
The SCU has the responsibility of deskewing Bus In
information except during read data transfers.

6.4.3 Register Definitions

BUS OUT

For extended instructions, NORMAL END is not


presented until after the read or write operation is
completed which is some undetermined time after TAG
GATE is reset. NORMAL END is reset by RESPONSE
after an extended operation.

BUS OUT
TAG OUT

II

TAG GATE
I
I
SELECT HOLDI
I

~,

II
100
NSEC

l- X
I

SYNC OUT

,I
--t
I

100
NSEC

\ .. MIN ..

l.I

DRIVE TAGS
-700 NSEC MIN
CONTROLLER TAGS
-100 NSEC MIN

Figure 6-26. SCU to Controller Interface Timing

'II

. MIN

_BU.;.,S_N_ _--!1125 NSEC


I

,..

MIN

~:..

S~YN~C~I~N_ _ _ _ _ _~I

100
NSEC

60 NSEC
..

I..

MIN .. 1

125 NSEC""'_ _ __
155 35 NSEC

MIN

..:..

..I

LI_ _ _ _ _ _ _ _ _ _ _ __

Figure 6-27. ContrqUer:. to SCU Interface


Timing (Data Transfers)

CO Register
The CO Register generates the tag out bits for the
Controller along with SELECT HOLD and CTL TAG GATE
for control of the file. The register can be loaded from
either the D Bus or the T Bus.

Check End
CHECK END indicates that an abnormal ending condition
exists. The abnormal condition is presented on Bus In
along with proper parity during the duration of CHECK
END. For Read or Write operations, CHECK END stays on
and BUS IN maintains proper parity until the Storage
Control acknowledges the receipt of the abnormal end
status information with the RESPONSE line. CHECK END
is used for extended operations only.

co
co

co

REGISTER BIT ASSIGNMENTS

co 2

SE
SELECT QUENCE TAG
BUS 0
OUT
OUT

co

TAG
BUS 3

co

TAG
BUS 4

co

TAG
BUS 5

co

TAG
BUS 6

co

TAG
BUS 7

The tag out bits (0, 3-7) have odd parity generated to be
sent with the tag bits to the Controller.

6-31

decremented with each SYNC IN tag from the


Controller. When the counter reaches zero, the
DTC c 0 branch condition IS set on.

DO Register
The DO Register is llsed when transferring data to the
file In the write mode. or used to define the command to
the Controller in conjunction with the tag out bits defined
by the CO Register. The register can be loaded from
either the 0 Bus or the T Bus.
DO .REGISTER BIT ASSIGNMENTS
DO 0

DO 1

DO 2

DO 3

DO 4

DO 5

DO 6

D07

BUS
OUT

BUS
OUT
1

BUS
OUT
2

BUS
OUT
3

BUS
OUT
4

BUS
OUT
5

BUS
OUT
6

BUS
OUT
7

Read Buffer Register


The Read Buffer (DI) Register is a one-byte specialpurpose register used to synchronize and buffer the read
data between the SCU Microprocessor and the Controller. Transfers of data between the 01 Register and
Controller are done one byte at a time with SYNC
IN/SYNC OUT control tags.
DI REGISTER BIT ASSIGNMENTS

During write operations, the DO Register is loaded by the


control program one byte at a time. The load of DO takes
place whenever the control program specifies the DO as
a destination on either the T or 0 Bus. There is a twobuffer sequence for all write operations in the SCU. It is
initiated by loading DO with the physical 10, and then
raising the Load Write Buffer (BUF/W) signal (RWC bit 0)
to gate DO to BUF/W. Then DO is loaded with the next
byte. Hence, the SCU has two bytes ready to transfer to
the Controller upon receiving SYNC IN.
Data Transfer Counter
The Data Transfer Counter (OTC) Register is a sixteen-bit
special-purpose, up/down counter register. The OTC
Register is loaded as a two-byte register from R2 and R3
on T Bus via the control statement, C(OTC = R23).
The OTC Register is loaded by the control program to
specify the number of data bytes to be transferred
between the SCU and the Controller. The OTC can also
be used as a special-purpose counter.
1. Read Operations
During the time that a read mode is indicated
(RWC bit 6 is set), the OTC Register is
decremented with SYNC IN. Since the DTC
Register is synchronized with the machine
clock, the OTC Register contents in the ECR
Register may be sampled at any time by use of
MCM bits 2 and 3.
2. Write Operations
During the time that a write mode IS indicated
(RWC bit 7 is set), the DTC Register is

DIO

Dll

DI2

DI3

DI4

DI5

DI6

DI7

BUS
IN
0

BUS
IN
1

BUS
IN

BUS
IN
3

BUS
IN
4

BUS
IN

BUS
IN

BUS
IN
7

During read operations the DI Register is loaded one byte .


at a time at a nominal rate of 1.24 microseconds per
byte. The actual transfer of the assembled byte takes
place with SYNC IN synchronized by a clock signal. When
DI is loaded, the OBR branch condition is raised to the
microprocessor. When DBR is raised, the control
program must initiate a read of DI within approximately
1000 nanoseconds or an overlay of the byte may take
place (430 nanoseconds is the worst case delay between
the load and rise of OBR plus control program branch
uncertainty). Thus, not counting the instruction to branch
on OBR, the control program has two instructions to read
DI. The read of DI can occur at the same time that the
CBR condition is set. Initially, data bytes are placed in 01
starting with the physical ID of the field, following
initiation of the read mode in the RWC register. Data
bytes are placed in 01 as long as RWC bit remains set
and SYNC IN is latched. After RWC bit 6 is reset, no more
bytes are placed into DI by SYNC IN until the read mode
is again initiated.

RWC Register Bit 0 Gate DO Reg to Write Buffer. This


bit loads the Write Buffer Register with the contents of
the DO Register. The contents are either a command to
the Controller or data to be written into the files by a
write command.

When bit 4 is high, the OTC Register will count UP with


each clock pulse. When bit 4 is low, the OTC Register will
count down with each clock pulse.
RWC Register Bit 6 Read Mode. This bit on indicates
that a Read operation is in progress, and enables control
and timing pulses for any Read operation sequence to
the Controller.

RWC Register Bit 1 Enable Recycle for Data Transfer.


This bit is used to enable the RECYCLE signal to the
Controller. RECYCLE is used to inform the Controller that
it should recycle its byte counter when it reaches zero
because there are more data bytes to be transferred
during read or write operations. The recycle function is
active as long as the DTC Register is greater than 15
during read or write mode.

RWC Register Bit 7 Write Mode. This bit on indicates


that a Write operation is in progress, and enables control
and timing pulses for any Write operation sequence to
the Controller.
ECR Register

RWC Register Bit 2 Gate Bus In. This bit is used to gate
data from the Read Buffer Register into the 01 Register at
Register Time during a Read Mode or command sequence. The data transfer takes place with or without
TAG VALID, NORMAL END, or CHECK END.

This register (Table 6-7) is pageable by bits 1, 2, and 3 of


the MCM register. The ECR register contains information
on the inbound tags from the Controller Check-2 error
conditions and the 16 bits of the OTC.
MCM Register

RWC Register Bit 3 Allow Response. This bit enables


the generation of the RESPONSE signal to the Controller
whenever NORMAL END or CHECK END is received from
the Controller in an extended operation.

The MCM Register is a four-bit register used to multiplex


the ECR Register contents to the Microprocessor. MCM
bit 4 with Latched Data Bit 6 (LDB 6) is used to define the
entries into the MCK + FOF multiplexor which is an input
to the A Bus of the ALU. The MCM Register is loaded
with data from 0 Bus bits 0-4.

RWC Register Bit 4 Not Used.


RWC Register Bit 5 Clock DTC Up. This bit controls the
incrementing or decrementing state of the OTC Register.

The D Bus bit 0 entry to the MCM Register is defined as


the "Enable CK2" signal, used to look for Check 2 errors.

TABLE 6-7. ECR BITS


RWC Register
The RWC register is used by the control program to
control the transfer functions of the CTL-1.

MCM
1 23

000

SELECT SELECT
ALERT 1 ACTIVE
CHECK

001

DI
BUFFER
CHECK

010
01 1

TAG
VALID

(NOT
USED)

SYNC IN

UNEXP.
END

SELECT
ACTIVE
CHECK

(NOT
USED)

CE
ALERT

DTC 0

DTC 1

DTC 2

DTC 3

DTC 4

DTC 5

DTC 6

DTC 7

DTC 8

DTC 9

DTC 10

DTC 11

DTC 12

DTC 13

DTC 14

DTC 15

NORMAL CHECK
END
END

RWC REGISTER BIT ASSIGNMENTS

GATE
DO TO
WRITE
BUFFER

CTL
RECYCLE

GATE
BUS
IN

ALLOW
RESPONSE

CLOCK
DTC
UP

CTLI
READ
MODE

CTL-I
WRITE
MODE

CHANNEL
INTERFACE
CHECKS

6.5 FE INTERFACE

~:~~~~EF~~~CK - - - - t l..~

INPUT CARDS

6.5.1 General
The FE (Field Engineer) interface contains the logic
required to manually address. load. start and stop the
SCU during maintenance procedures. The logic required
to control and display diagnostic routines and errors.
while in the In-line or Off-line mode. is also contained in
the FE Interface

FROM
MICR O
PROCESS OR

TIMING

ERROR
LIGHT
DRIVERS

..._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _..~~

VCf>02

The functional description of the FE Interface includes a


brief description of each functional block In the block
diagram of Figure 628

DISPLAY
SELECTION
LOGIC
VC502

ROLLER BAR
INPUT CMDS
FROM FE

DIAGNOSTIC
CONTROL INDICATOR
SIGNALS TO FE

SELO.1.2.3
SIGNALS TO

I-SWITCH
DRIVERS

SW O-SW 7

es o-es

VC501

DATA. PARITY
DISPLAY
STROBES
SEL 0 1 2 3

PRESENT/LAST
ADDRESS
REGISTER

f'RESENl LAST ADDRESS

VD500'

LOAD START
ADDR.
SEL 0 1 2 3

The Diagnostic Indicator Driver and MD2 Register logic


buffers. drives and gates the Diagnostic Control Indicator
signals from the Microprocessor. via the 0 Bus. to the
DIAGNOSTIC CONTROL indicator on the FE panel

START
ADDRESS
REGISTER

START ADDR

VD501'

The Display Selection Logic gates microprocessor timing


pulses and uncoded control panel commands to generate
data and parity display strobes for the Display Register.
The logic also generates select enable signals to the
various input boards and to the Address Register Control
logic.

LOAD A.
SEL 0 1 23

STOP A
ADDRESS
REGISTER

--

LOAD B
SEL 0 1 23

STOP B
ADDRESS
REGISTER
VD500-

MACHINE:
CONTROLS
FROM FE

TOGGLE
SWITCH
LOGIC
ILATCHESi
VD502
VC504

VD503'

The Start Address register stores and gates the LOAD


ADDRESS Input commands from the FE panel. and is
strobed by a LOAD START ADDR command
The Stop A Address Register stores and gates the LOAD
ADDRESS Input commands from the FE panel. and is
strobed by a LOAD A command

The Data Display Register stores up to 36 bits of data


which are displayed on the FE panel Tlte register IS
strobed by a Data Set Strobe pulse. or by a Parity Set
Strobe pulse
The Address Comparison Circuits Input address bits from
the Start Stop A and Stop B Address Registers and
compare the bits with the address bits from the
Microprocessor Each Address Register control circuit
generates a pulse (SYNC EQUAL STOP A EQUAL. STOP
B EQUAL) whenever compartson IS attained

For FE panel operations requirtng access to the WCS


expanSion feature. a thirteenth address bit (bit 12) IS
provided to access locations greater than 4095 (OFFF
hex) The switch and present/last indicators associated
with thiS address bit are located on the EX board. along
With the sync/stop A/stop B equal logiC

ADDRESS
COMPARISON
CIRCUITS

STOP B ADDR

,
ADDR BITS F ROM
MICROPROCE SSOR

DATA BIT DISPLA y


SIGNALS TO FE

-ADORE SS BIT 12 113TH BIT)


<UNCTI ONS ARE LOCATED
ON EX PCB. USED FOR FE
PANEL OPERATIONS WITH
WCS ADDRESSES >FFF
I

The Switch Drivers drive command signals from the


Roller Bar switch to the ALU.

LJISPLA'
REGISTER

STOP A ADDR

VD500'

The Present/Last Address Register buffers the address


bits from the Microprocessor and gates them with the
uncoded Control Panel Display Commands for display on
the FE panel

The Stop B Address Register stores and gates the LOAD


ADDRESS input commands from the FE panel and .s
strobed by a LOAD B command

VD

The Error Light Driver buffers and drives the MACHINE


CHECK and DIAGNOSTIC CONTROL signals from the
various printed circuit boards (i.e .. GE. GO. CM. etc.) to
be displayed on the FE panel.

The Roller Bar Switch Interface gates and drtves


MACHINE CONTROL commands to the Input cards and to
the Address Register Control logic and comparison
Circuits

TO INPUT
CARDS

CS O-CS 7
SWO-SW 7

ROLLER BAR
SWITCH
INTERFACE

The Toggle Switch Logic contains the latches and gating


for the MACHINE CONTROL toggle switches on the FE
panel.

DIAGNOSTIC
INDICATOR
DRIVER AND
MD 2 REGISTER
VC

It. PUT CARDS

6.5.2 Functional Description

MACHINE CHECK
ERROR SIGNALS
TO FE

IID500-

LOAD ADDRESS
INPUTS FROM FE

TOGGLE
SWITCH
LOGIC
(GATING)

cs O. CS 2
TO INPUT CARDS

VC502
VC504

Figure 6 28 FE Interface Block Diagram

6-33

6.6 CHECK-1 ERRORS

6.6.1 Error Detection and Sequences

and the channel issues a Selective Reset (raises SUPPRESS OUT and drops OPERATIONAL OUT). If
OPERATIONAL IN is down when the error is detected,
the SCU raises REQUEST IN. Then in response to
COMMAND OUT In the SCU-initiated sequence. the SCU
raises DISCONNECT IN and the channel follows with a
Selective Reset.

The 3672 Storage Control Unit includes circuitry which


detects the presence of internal Microprocessor errors
(called Check-l errors). These error conditions are
catastrophic in the sense that the SCU functional
microprogram cannot recover from them with any
assurance of proper microprogram orientation or
cl!stomer data integrity. (Check-l errors are in contrast to
Check-2 errors, which are recoverable.) For this reason.
the detection of a Check-l error forces a program halt
and the initiation of a channel sequence which, in turn,
forces the channel to issue a Selective Reset. The
Selective Reset restarts the microprogram. The
microprogram then resets its operational status and
assembles the error information which was stored in the
Machine Check (MCK) Register upon error detection. This
error information is sent to the channel upon a Sense
I/O command following the next Test I/O command.

6.6.2 Machine Check (MCK) Register

The channel sequence following Check-l error detection


falls into two categories, depending on the state of the
OPERATIONAL IN signal. If OPERATIONAL IN is up when
the error is detected. the SCU raises DISCONNECT IN

The MCK Register consists of 48 bits which contain


information concerning the error status of the machine.
These bits are multiplexed to the A Bus by MCM bits 5, 6
and 7 as shown in Table 6-3 of paragraph 6.3.4.2. Error
bits
through 15 are the actual Check-l error bits. If all

A DISCONNECT IN interlock in the SCU channel interface


prohibits the raising of DISCONNECT IN after a Selective
Reset until the point where ADDRESS IN is raised in the
next channel sequence. If a Check-l error occurs while
DISCONNECT IN interlock is active, the microprogram
stops and the TAGS IN lines are frozen. At this point. the
channel can ignore the SCU or. after a time out, issue a
reset. The DISCONNECT IN interlock circuitry insures
that the channel does not get into a DISCONNECT IN
Selective Reset loop.

these bits are false. the SCU Microprocessor is functioning properly. When anyone of these error bits is set,
the status of all 16 bits (0-15) is frozen as well as the
failing address (27-39) associated with the error(s). If the
active error bit is MULTIPLE WCS ERROR (1), the failing
WCS error-correcting parity pattern is. also frozen (1623). If the active error bit is A Bus, B Bus or T Bus parity
error (2-4), then MCK bits 40 through 43 point to the PCB
(GA GB, GC or GO) which contains the source register
associated with that particular parity error. MCK bits 25,
26 and 44 through 47 are Check-2 errors.

6.6.4 Reset of Check-1 Error Conditions

6.6.3 Check-1 Error Display

Although the SCU Check-l error circuitry will stop the


clock each time an error is detected, the state of the MCK
Check-l error bits is frozen once any Check-l bit is set
(MCK bits 0-15). To arm these MCK bits for further error
latch-up, these bits must be reset. The microprogram can
issue a C(RESET.CKS) control statement to perform this
function.
Under normal
system
operation,
the
microprogram will reset the Check-l bits after it has
assembled the error status into the sense bytes following
a Selective. Reset. In addition, the FE can reset the
Check-l bits and failing address using the FE RESET
switch. As a further control option, the FE can allow the
microprogram to continue to run when a Check-l error
occurs through use of the CHECK-l OVERRIDE switch.

When a Check-l error is detected, one or more discrete


LED indicators on the SCU FE panel illuminate, defining
the particular error detected. The state of these indicators is derived directly from MCK bits 1 through 15
and 40 through 43. For example, the BRNCH error
indicator is illuminated if MCK bit 5, 8 or 11 is set
(Branch Error, BR Multiplexer Error, or Sub-routine
Error). The indicator associated with each MCK bit is
listed with the MCK bit definitions in Table 6-8. If there is
a Check-l error indicated by the LED matrix, then the
address displayed in roller position E of the FE panel is
the failing address associated with that error.

When the SCU is in the Scan mode, all Check-l error


latches are held reset, except MULTIPLE WCS ERROR
and WCS ECC ERROR. If a noncorrectable WCS error
occurs, the clock stops with the appropriate error indicator (RAM or ECC) illuminated and the failing address
displayable at roller position E. The FE RESET switch
resets these error bits if the SCAN switch is deactivated.
The CHECK-l OVERRIDE switch can also' be utilized
while in Scan mode. The only Check-2 error enabled to
stop the clock while in Scan mode is SINGLE RAM
ERROR. No indicator will be displayed except STOP
CLOCK. This error can also be degated.

Table6-B. MCK REGISTER DEFINITION

BIT

DEFINITION
CHC PARITY ERROR
CHC Register contains wrong parity error
MULTIPLE WCS ERROR
A
Illuiliple (noncorrectablel

WCS

ERROR
CLASS

DISCRETE
DISPLAY

CKl

None

CKt

RAM
Indicator

accessing

A BUS PARITY ERROR


A Illiscompare of the parity generated for a
register on A Bus and the parity bit previously
stored for that register

CKl

B BUS PARITY ERROR


A mis.:;ornpare of the parity generated for a
register on B Bus and the parity bit previously
stored for that n:gister

CKl

T BUS PARITY ERROR

CKl

T Bus
Indicator

BRANCH ERROR
a Miscompare of address parity and incrementer
parity if increment cycle.
b
Miscompare of address parity and SR parity if
B (ADR = SRI cycle.
c. Detection of simultaneous Branch Set and
Increment Set signals.

CK-l

BRNCH
Indicator

IMPL ERROR
During hardware IMPl. one of the following
conditions was detected:
a. Single Track Seek was not completed within
one second.
b A sync byte was not found in the FD read
data within one second.
Read overrun.
Sixteen occurrences of:
1 No TRACK = 0 decode (after TRACK 0 line
from FD valid). or
2. Parity error detected on FD Read Data

CKl

ALU ERROR
a. Miscompare of duplicate ALU outputs including
D = FF and carry conditions.
b. Miscompare of latched 0 Bus with the
ALU output.

CKl

BIT

A Bus
Indicator
10

A miscompare of the parity generated for a


register on T 8us and the panty bit(s) previously

B Bus
Indicator

ERROR
CLASS

DISCRETE
DISPLAY

BR MPXR ERROR
Miscompare of parity generated for a register
at the output of the BR Multiplexer
with the parity bit previously stored for that
register.

CKl

BRNCH
Indicator

PS ERROR
Miscompare of duplicate PS registers.

CKl

PAGE
Indicator

CKl

TRANS
Indicator

DEFINITION

OP CODE TRANSLATION ERROR


Invalid control field status
OP code.

for

DISCRETE
DISPLAY

DEFINITION

26

SINGLE WCS ERROR


Single (correctable) WCS accessing error

CK2
(in SCAN
only)

FAILING ADDRESS 120


Address to be associated With the Checkl
error(s) latched in MCK bits 0 through 15

CKl

40

GA PCB
The PCB containing the source register for an
A Bus. B Bus, or T Bus panty error IS the
GA PCB.

CKl

GA
Indicator

2739

the

ERROR
CLASS

BIT

Roller
position

11

SUBROUTINE ERROR
Miscompare of SR parity and address parity
parity for C(LD.SR.+I) or B(LD.SF.+I).

CKl

BRNCH
Indicator

41

GB PCB
The PCB containing the source register for an
A Bus parity error is the GD PCB.

CKl

GB
Indicator

12

WCS ECC ERROR


Invalid combination of overall WCS access parity,
corrected word parity, and any P.

CKl

ECC
Indicator

42

GC PCB
. The PCB containing the source register for an
A Bus or T Bus parity error is the GC PCB.

CKl

GC
Indicator

13

MULTIPLE BUFFER ERROR


A multiple (noncorrectable)

CKl

BUFF
Indicator

43

GD PCB
The PCB containing the source register for an
A Bus parity error is the GO PCB.

CKl

GO
Indicator

CKl

BUFF
Indicator

44

FD READ ERROR

CK2 or
CK-l (if
IMPL)

stored for that registei.

14

buffer

BUFFER ECC ERROR


Invalid combination of buffer
corrected word parity, and any P.

accessing

access

parity.

15

BUFFER WRITE T BUS ERROR


Miscompare of T Bus parity with parity
generated at the input of the buffer prior to a
write. C(BUF = R03).

CKl

16.23

FAILING Pl-PS
The a-bit error correcting parity pattern associated
with a multiple or single WCS error (MCK 1
OR MCK26).

CK-lor
CK-2

24

SELECTIVE RESET
Indicates that channel has issued a Selective
Reset (Microprogram checks this bit in the reset
procedure to distinguish between General and
Selective Reset).

25

SINGLE BUFFER ERROR


Single (correctable) buffer accessing error.

IMPL
Indicator

ALU
Indicator

a.
b.
c.

BUFF
Indicator

45

CK2
(for diagnostic
purposes only)

Parity error was detected on the one


second, or
Read overrun, or
A sync byte was not detected in the FO
read data within one second.

FD SEEK ERROR
a. A Seek was not completed within one
second. or
b. A TRACK
0 Read decode was not found
within two sectors after the TRACK 0 line
from the FD went active.

46

FD NOT READY
A time delay signal which should be 57
seconds in duration after FD power is initiated

47

FD WRITE BUS ERROR


A parity error was detected at the
FDC registers.

CK2 or
CK-l (if
IMPL)

CK2
FDD

or

NOTE:
1111 MCK Ill!" co111ll! displayed through SET DISPLAY of MCK register (microprogram display).

6-35

GLOSSARY OF TERMS

ADDR. ADRS
ALU
AM

Address
Arithmetic Logic Unit
Address Mark

B
BAR
BI
BO

Branch
Buffer Address Register
Bus In
Bus Out
I/O Instruction Base Address
Register Location

B1
C
CAR
CAW
CBR
CC

Count
Cylinder Address Register
Channel Address Word
Channel Buffer Ready
Chain Command. Channel Byte Count.
Condition Code

C.C
CD
CCW
CE
CHC
CHF
CHNL
CI
CK
CKD
CLK
CMD
CP

Cylinder Address (two bytes)


Chain Data
Channel Command Word
Channel End
Channel Control
Channel Flags
Channel
Channel Interrupt
Check
Count/Key/Data
Clock
Command
Channel Priority

3672.21-0001-10/75

CPU
CT
CTL
CTL-I
CUDI

Central Processing Unit


Channel Tag
Controller
Controller Interface
Control Unit Device Interface

0
DC
DCC
DDM
DE
DEV
01
DL
DO
DTC
0
1

Data
Data Counter
Disconnected Command Chaining
Disc Drive Module
Device End
Device
Device Interrupt
Data Length
Data Out
Data Transfer Counter
I/O Instruction Displacement

ECC
EOB
EREP
ERP

Error Correction Code


End of Block
Error Recovery Procedure
Error Report

FO
FOC
FDD
FE
FIDS
FLO
FSR

Flexible Disc
Flexible Disc Control
Flexible Disc Drive
Field Engineer
Fault Isolation Diagnostic System
Field
Field Service Representative

SCTR
SCU
SEQ
SERDES
SLI
SR
SW

Sector
Storage Control Unit
Sequence
Serializer /Deserializer
Suppress Length Indicator
Subroutine
Switch

R
RN
RAM
RST
R/W
RWC
R03

Record
Record n
Random Access Memory
Reset
Read/Write
Read/Write Control
Registers 0 through 3

No Operation

TIC
TR
TRK

Transfer In Channel
Transmit Request
Track

OLTEP
OP

On Line Test Executive Program


Operation

UC
UE

Unit Check
Unit Exception

P
PCI
PLO
PS

Problem State
Program Control Interrupt
Phase Locked Oscillator
Page Select

VFO

Variable Frequency Oscillator

W
WCS

Wait State
Writable Control Storage

QE

Queue Empty

J.L

Micro (used as prefix. as in J.Lprogram)

Gn
HA
HAR
H.H

Gap n
Home Address
Home Address Register
Head Address (two bytes)

10
ILC
INCR
I(M)PL

Identifier
Instruction Length Code
Increment
Initial (Micro) Program Load

K
KL

Key (or Constant = 1024)


Key Length

M
MCK
MPXR
MUX

Machine Check Mask


Machine Check
Multiplexer
Multiplexer

NO-OP

G-1

COMMENTS FORM

3672 Storage Control Unit Theory of Operations -

3672.21-00/0001

Pleas'e send us your comments; they will help us produce better publications.
Whenever possible, please use specific page and paragraph references.

Is the material:
. Easy to understand?

Yes

No

o
o
o
o
o

Conveniently organized?

Complete?

o
o
o
o

Well illustrated?
Accurate?
Suitable for its intended audience?

For what purpose did you use this publication (reference, general interest, etc.)?

Please state your department's function:

Please use the space below for specific comments which you feel would improve any
aspect of this publication. All comments become the property of Memorex Corporation,

~.

First Class

Permit No. 250


Santa Clara
CaHfornia
95050

Business Reply Mail


No Postage Necessary if Mailed in the United States

Postage Will Be Paid By

d]J

Memorex Corporation

Publications Department
Mail Stop 14-07
San Tomas at Central Expressway
Santa Clara, California 95052

Thank you for your information ......... .

Our goal IS to provide better, more useful manuals, and your


comments will help us to do so .

.... ..... .Memorex Publications

You might also like